TokenmaxxingAI token costsAgents online
GuidesLabsEventsLeaderboardToolsBriefing
tokenmaxxingMAINLINE
Join Mainline
← @drewx_designchangesshow all projects

Changes behind the numbers

Every change that reached a project's default branch. Greyed rows were not counted, with the reason. A project from an older card may show dates and line counts only.

Grantsville

LIVE1410 landed · 1189 itemized
  • chore: declare the production homepage in package.json (#2549)#25491 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(roadmap): excerpt long mandatory bodies instead of refusing, and raise the …#2545289 lines
  • fix(roadmap): raise the ordinary Now cap from 3 to 8 to match the milestone spe…#2541258 lines
  • test(ci): register docs/audits/seo-weekly as a mentions-only reader path (#2548)#25481 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(roadmap): carry the website-counters audit and candidate sidecar to main …#2546104 lines
  • docs(seo): land the 2026-10-22 weekly SEO audit and its outcome candidates (#25…#2544227 lines
  • fix(scheduled-runner): five-start cold-start respawn budget; postscript gets fo…#2535103 lines
  • fix(deploy): re-fetch unsettled CI job evidence before refusing qualification (…#2540291 linesfolded into the change before it
  • ci: raise test-serial-risk-1/2 timeout-minutes from 20 to 25 (#2539)#253920 lines
  • chore(roadmap): rehash the Project views follow-up sidecar; record the assessme…#2534304 lines
  • test(ci): launch workflow-contract bash outside the fixture dir (Windows EBUSY)…#2533112 lines
  • fix(seo-agent): give the agent repo SHAs, allow Railway whoami, and bound the l…#2523919 lines
  • feat(resources): show when each guide was last reviewed and list guides in llms…#2519262 lines
  • feat(extraction): field decision table and de-abstention in the faithful progra…#2511472 lines
  • fix(extraction): close the four faithfulness-guard accept-side gaps (#2460, PR …#2526859 lines
  • fix(roadmap): GLM seat breaker ignores pre-launch input refusals (#2530)#2530224 lines
  • docs(ci): record the #2467 smoke measurement result (target met) (#2529)#25292 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(scheduled-runner): retry cold-start pending connectors, pin PATHEXT, decode…#2513586 lines
  • fix(roadmap-intake): share one filing census across worktrees with reuse checks…#25144214 lines
  • fix(extraction): headings, paragraphs and table cells are faithfulness-guard un…#2507579 lines
  • fix(worktree-cleanup): guard each snapshot file cheaply instead of rechecking t…#2527397 linesfolded into the change before it
  • fix(sorter): reuse whole sources whose producer snippet is unchanged at main (#…#2525316 lines
  • feat(resources): lead the funder-access guide into the qualification screen (#2…#2516118 lines
  • feat(marketing): publish a dated AI-use position page at /ai-use (#2518)#2518452 lines
  • feat(grant-monitors): local Grants.gov job, fetcher, and shared Grant Opportuni…#24814179 lines
  • fix(extraction): handle thin extractions explicitly across all readers (#2508)#25083486 lines
  • docs(roadmap): Horizon Board and flow-ordered Project 1 views (#2506) (#2509)#2509429 lines
  • fix(roadmap): hourly sorter compiles again by reusing verified-unchanged eviden…#25052977 lines
  • feat(seo-agent): run the weekly SEO agent locally through a confined scheduled …#249914408 lines
  • docs(discovery): roadmap intake files audit findings that main has already fixe…#2504179 lines
  • docs(discovery): Grantsville Roadmap Project views hide the Horizon flow (#2503)#2503153 lines
  • docs(discovery): roadmap filing recaptures the whole-board census on every atte…#2501202 lines
  • docs(audits): launch quality bar gaps (#2493, #2494) (#2498)#2498197 lines
  • docs(discovery): worktree cleanup removal gate reruns a full recheck twice per …#2497159 lines
  • docs(discovery): record the two commission retry findings fixed by #2487 (#2491)#2491276 lines
  • fix(worktree-cleanup): keep the recovery-check reason and let a supervised conf…#24871091 lines
  • perf(roadmap): cache verified immutable artifacts per invocation; status 152 s …#2490506 lines
  • fix(roadmap): raise sorter wrapper status/delivery limits to 900 s (stopgap for…#248856 lines
  • docs(research): correct grant-sources figures against their own data (#2489)#248933 lines
  • docs(research): where small nonprofits can get non-foundation grant money (#248…#24861311 lines
  • fix(roadmap): admit turn/moderationMetadata in Astra assessments; scheduled sor…#2479688 lines
  • docs(content): borrow seven content and SEO moves from Kindora's site (#2478)#2478485 lines
  • docs(strategy): mark the qualification screen live, land the September competit…#2476418 lines
  • chore(roadmap): record Drew's Type and Priority on three filed-pending candidat…#247442 lines
  • docs(research): Kindora website teardown and three takeaways for Grantsville (#…#2471410 lines
  • docs(discovery): a backslash path in a tracked brief can fail the Tailwind buil…#2418110 lines
  • docs(discovery): worktree-cleanup UNC colon-param and hex-key entropy gaps (not…#2470286 lines
  • perf(ci): keep TypeScript incremental state between smoke runs (#2458) (#2467)#246736 lines
  • docs(discovery): record low prompt cache hit rate for later brainstorm (#2447)#2447138 lines
  • docs(audits): three roadmap sorter follow-ups left after the F-LIVE fixes (#246…#2466254 lines
  • chore(roadmap): rehash the intake-lock discovery candidate's cited sources (#24…#24656 lines
  • fix(roadmap): sorter live-run findings F-LIVE-1/2/3 (#2464)#2464648 lines
  • docs(audits): smoke lane cap headroom finding (#2458) (#2459)#2459132 lines
  • feat(ci): main reuses the queue's DB-free lanes when bound (#2394 phase 2) (#24…#24461586 lines
  • fix(roadmap): compact the assessment evidence view so canonical filing fits the…#2444562 lines
  • docs(ci): record the #2394 queue binding shadow report (U4) (#2445)#2445113 lines
  • feat(roadmap): read-only snapshot rehearsal through the live transport (sorter …#24431826 lines
  • docs(audits): agent-ops CLI test LOI run 2 findings and 9 outcome candidates (#…#2442739 lines
  • test(roadmap): make the never-assessed replacement-truncation test fast (#2441)#2441214 lines
  • feat(roadmap): snapshot executor scheduling with classification plans (sorter B…#24352027 lines
  • feat(roadmap): census-free anchored target observations (sorter B12 Unit 4) (#2…#24341125 lines
  • test(roadmap): make the batch-approve chunking tests fast (queue-blocking timeo…#2437115 lines
  • feat(roadmap): snapshot plan and issue-skip journal events (B11/B12 Unit 2) (#2…#2432822 lines
  • fix(worktree-cleanup): stop entropy secret rules vetoing code and table prose (…#2426221 lines
  • feat(roadmap): batch observations and a chunked approve re-collect (B11 Unit 3)…#24331533 lines
  • test(worktree-cleanup): cut Windows cleanup test setup spawns (#2423) (#2428)#2428221 lines
  • feat(roadmap): snapshot transition contract for the B11/B12 sorter redesign (Un…#24311660 linesfolded into the change before it
  • fix(roadmap): GraphQL pagination fails closed on hasNextPage without a cursor (…#2430365 lines
  • fix(worktree-cleanup): count PowerShell -Param:<tree> as transcript activity (#…#242738 linesfolded into the change before it
  • docs(audits): three worktree-cleanup activity-reader shape gaps (A-C, not filed…#2429340 lines
  • fix(ci): keep drafts off the Windows cleanup lane and shard it under its cap (#…#2421214 lines
  • ci(rules): give the rules lane an 8-minute variance margin (#2412)#241213 lines
  • fix(worktree-cleanup): exempt a release call only when it runs node, git or gh …#242520 linesfolded into the change before it
  • fix(worktree-cleanup): one missed nightly no longer blocks owner release (#2406…#2424119 lines
  • feat(vault): register prog.plans_and_targets dark behind the faithful contract …#237410936 lines
  • feat(ci): record main's same-SHA merge-queue binding in shadow mode (#2394) (#2…#24151378 lines
  • test(ci): fix the reader-guard and popover-label merge-queue timeouts (#2393) (…#2408173 lines
  • docs(audits): two worktree-cleanup activity-reader gaps from the #2413 review (…#2416233 linesfolded into the change before it
  • fix(roadmap): a seat outage no longer erases settled local-path coverage (#2417)#2417781 lines
  • fix(worktree-cleanup): admit placeholder URL creds at example hosts (#2414)#2414118 lines
  • fix(worktree-cleanup): stop a plain owner-release call from holding its own tar…#2413506 lines
  • fix(roadmap): a moved timeline cursor no longer voids an approved snapshot (#24…#2401194 lines
  • fix(roadmap): an open linked PR's head push no longer voids an approved snapsho…#2400300 linesfolded into the change before it
  • feat(llm): move remaining haiku-4.5 sites to haiku-5.5 behind pre-registered ch…#23973092 lines
  • feat(roadmap): behavior-neutral phase timing and rate-limit telemetry for live …#2398816 lines
  • docs(audits): high-entropy secret rules veto prose and code in roadmap data (#2…#2407172 lines
  • fix(roadmap): bound the sorter's final stdout print below the wrapper cap (#239…#2396621 lines
  • docs(audits): worktree cleanup findings from the 2026-10-15 nightly (#2402)#2402364 lines
  • chore(claude-md): review CM-22 and set its next review date (#2403)#24032 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(roadmap): accept a journaled same-material recapture at the snapshot gate (…#2395113 lines
  • feat(agent-ops): U20b close the last deferred parity rows (4506-B, reconciliati…#2388971 lines
  • feat(worktree-cleanup): snapshot scratch-only worktrees to rescue refs before r…#23925359 lines
  • feat(agent-ops): U20a close deferred parity rows; agents can no longer lock fac…#23841300 lines
  • fix(roadmap): register policy v5 with a 3-hour run deadline (#2391)#2391151 lines
  • fix(ci): docs-only merge-queue entries skip the heavy lanes (#2390)#2390477 lines
  • fix(worktree-cleanup): give the admission identity probe its own 10 s bound (#2…#2386142 lines
  • feat(roadmap): outcome dispositions in the outcome-candidate envelope (#2377)#23777498 lines
  • fix(ci): load only scannable files in the analytics parity lint (#2383)#2383114 lines
  • docs(discovery): in-app vault write and swap tools bypass the agent write guard…#2382116 lines
  • fix(sorter): tests use an injected host-lock port, production stays fixed (#238…#2381336 lines
  • feat(agent-ops): Phase 4 vault and voice — vault reads and agent-provenance wri…#23709571 lines
  • docs(discovery): apply accepts decision-report fingerprints outside the report'…#2375131 lines
  • feat(sorter): a GLM local-path refusal settles coverage as single-seat glm_refu…#2372657 lines
  • fix(ci): treat the CI runbook as docs now that no code reads it (#2379)#237940 lines
  • fix(sorter): census pages of 50 and a bounded retry for GitHub's untyped GraphQ…#2373369 lines
  • test(autopilot): run static reproduce-E2E refusals in-process (#2364) (#2368)#2368715 lines
  • docs(discovery): create-reconciliation tests take the real intake lock (#2371)#2371135 lines
  • docs(plans): worktree cleanup scratch snapshots plan (#2369)#2369561 lines
  • feat(sorter): record a sanitized diagnostic for every failed gh read (#2367)#2367500 lines
  • fix(sorter): GLM seat checks decoded prompt content for local paths (#2366)#2366217 lines
  • fix(worktree-cleanup): remove dangling-junction and Next self-link trees (#2353…#23621350 lines
  • docs(discovery): agent-ops voice.update can stamp confirmed voice state without…#2363110 lines
  • feat(sorter): Astra seat runs a sorter-owned, pinned Codex 0.160.0 copy (#2295)…#23612053 lines
  • feat(agent-ops): Phase 4 grant reads and recovery — grant.read includes, grant.…#236011510 lines
  • fix(sorter): never-assessed members first in the shadow new-filing fallback (#2…#235986 lines
  • docs(ci): record adopted budget policy and reuse decision (#2358)#2358175 lines
  • fix(ci): report demand and cost gaps explicitly (U2) (#2357)#2357621 lines
  • perf(ci): reduce registration-ratchet check time (U4) (#2356)#2356133 linesfolded into the change before it
  • Roadmap shadow observations (#2086)#2086187568 lines
  • <redacted>#233480 linesfolded into the change before it
  • docs(inngest): resync after the deploy is live; record the budget guard's step-…#2102200 linesfolded into the change before it
  • docs(plan): worktree cleanup dangling junctions and build self-links (#2353)#2353449 linesfolded into the change before it
  • docs(audits): CLI test LOI findings (#2341, #2342) (#2343)#2343183 linesfolded into the change before it
  • Rescued worktree documents (#2071)#20713397 linesfolded into the change before it
  • docs(content): make demand measurement a gate, and record the failure modes (#2…#2089108 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan): evidence-gated outcome dispositions in the roadmap envelope (#2347)#2347357 linesfolded into the change before it
  • feat(agent-ops): Phase 4 structure and planning — item.structure, plan.edit, se…#235010427 linesfolded into the change before it
  • ci(worktree-cleanup): give the Windows lane 45 minutes (#2355)#23552 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): opt-in v2 lifecycle-receipts export with per-seat provenance (so…#2351714 lines
  • fix(roadmap): recognize Z.AI plan usage limits and stop a slot that keeps faili…#2354267 linesfolded into the change before it
  • fix(worktree-cleanup): release Codex leases held by reused PIDs; trim and time …#2352946 lines
  • docs(plan): graph-order criteria authorship and the reversion note for the clos…#234928 linesfolded into the change before it
  • docs(plan): worktree cleanup lease identity and recheck trim (#2348)#2348409 lines
  • docs(plan): filing-side Lifecycle closer for sorter-flagged issues (#2345)#2345752 lines
  • feat(roadmap): attribute every GLM seat call and select its credential slot (so…#2346631 lines
  • fix(worktree-cleanup): remove worktrees under registration churn; ordered admis…#23402935 linesfolded into the change before it
  • feat(agent-ops): Phase 4 inputs — Drive folder, reacquire, reclassify, funder.q…#233819389 lines
  • fix(roadmap): pin each sorter run to the fetched main and prove ancestry at gen…#2339809 lines
  • feat(roadmap): recheck only changed, expiring or deferred issues in daily runs …#2335882 lines
  • feat(vault): segment-citation resolver and evidence gate; U6b result STOP (#218…#232115697 lines
  • docs(plan): extraction faithfulness for #2328 and #2296 (#2337)#2337617 lines
  • fix(roadmap): give each Astra assessment its own thread; invalidate shared-thre…#2336981 lines
  • feat(agent-ops): document list, read and update operations (Phase 4 PR9, U23) (…#23335287 lines
  • feat(drafts): list citations follow their entry, phase 2 (#2212) (#2326)#23263510 lines
  • fix(worktree-cleanup): give the nightly run 75 minutes (#2331)#2331207 lines
  • docs(plan): worktree cleanup under registration churn (#2332)#2332317 lines
  • fix(llm): count BYOK spend and record failure cost from the response (#2329)#2329575 lines
  • docs(audits): extraction restatements add or drop claims the source does not su…#2327119 lines
  • refactor(vault): route every evidence reader through readAtomEvidence (#2186 ev…#23181017 lines
  • docs(solutions): a hook fix reaches each worktree only after it merges main (#2…#214762 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(llm): record tokens and cost for billed generateObject failures (#2325)#2325233 lines
  • fix(documents): transcription canonical text keeps table rows, cells and list i…#2324310 lines
  • feat(agent-ops): one revision rule for the review panel, section.revise and the…#23074841 lines
  • feat(roadmap): assess two issues at once under the shared global budget (#2315)#2315822 linesfolded into the change before it
  • fix(worktree-cleanup): walk a lockless node_modules; record phase durations (#2…#2323514 linesfolded into the change before it
  • feat(vault): evidence writer, read predicate, sole-access lint and legacy value…#23171060 lines
  • fix(roadmap): match stateReason across GraphQL and REST casing (unblocks the sh…#231919 lines
  • docs(audits): reproduce auto-fix calibration close-out and pause decision (#232…#2322111 lines
  • perf(ci): run three resolving AST gates on one shared ts-morph program (#2313)#2313601 lines
  • fix(autopilot): observe enforces the check rules that reach the folded symptom …#2314974 lines
  • feat(vault): source evidence persistence contract (#2186 evidence U1) (#2310)#231024884 lines
  • fix(roadmap): spend projection reads only when they can change the compile (#23…#2311435 linesfolded into the change before it
  • docs(audits): resolving one twin flag hides the other while it still blocks rea…#2292110 lines
  • feat(documents): canonical-text converters and document segmenter (#2186 eviden…#23122150 lines
  • docs(audit): agent-ops Phase 3 production smoke findings (#2176)#2176197 lines
  • fix(roadmap): withhold unsafe repository excerpts instead of deferring issues, …#2309869 lines
  • docs(audits): list-record citations can repoint silently and miss the out-of-da…#2208317 lines
  • feat(vault): stable ids for list entries, phase 1 (#2212) (#2282)#228217225 lines
  • feat(roadmap): open-time start command recommends the skill to start an issue w…#23081129 lines
  • fix(worktree-cleanup): owner releases stop refusing links-unknown (#2300)#2300373 lines
  • fix(review): a resolved twin no longer hides a sibling that still blocks readin…#2306153 lines
  • fix(roadmap): accept filing's Effort receipts and name the collector failure (#…#2302188 lines
  • docs(audits): extracted plans and targets have no home (#2294)#2294108 lines
  • docs(plans): consolidate Phase 4b operations and add agent-surface rules (#2305)#230570 lines
  • feat(roadmap): report a recommended start command per issue in every sorter com…#2269918 lines
  • feat(autopilot): brief/validator parity contract for the reproduce stages (#229…#2299898 lines
  • docs(plan): source evidence foundation for #2186 (#2304)#23041093 lines
  • fix(agent-ops): give each twin flag its own address in flag.act and grant.read …#2301544 lines
  • docs(plans): agent-ops Phase 4 amendment — shared revision rules (U16) and defe…#2298348 lines
  • docs(roadmap): filing-alignment outcome dispositions; sidecar keeps only the ap…#2297159 lines
  • feat(agent-ops): Phase 4 vault atoms, voice profile and programs (#2278)#22788061 lines
  • fix(roadmap): re-pin the sorter's Astra seat on codex-cli 0.160.0 and run shado…#22901009 linesfolded into the change before it
  • fix(autopilot): the probe brief states every rule observe enforces (#2289)#228962 lines
  • fix(fonts): self-host Literata and DM Sans so the build never fetches Google Fo…#2285212 lines
  • feat(roadmap): filing records an authored Start command (#2273)#2273365 lines
  • feat(vault): directors confirm a fact's kind and answer the overstatement card …#22792926 lines
  • ci(vitest): split the serial-risk shards by measured duration (#2286)#2286245 lines
  • docs(audits): agent-ops flagRef collides for twin flags (#2281)#2281109 lines
  • fix(vault): stop compound writes and entry removals failing the passage-rewrite…#2284231 lines
  • feat(drafts): advisory warning when a draft states a planned or estimated fact …#22613860 lines
  • feat(agent-ops): Phase 4 internal-only package preview (#2280)#22802086 lines
  • fix(drafts): a unit paraphrase that names one element of a record_array no long…#2274464 lines
  • fix(roadmap): accept filing's Start segment on the Intended classification line…#2272100 linesfolded into the change before it
  • fix(autopilot): make the reproduce pipeline usable under mock policy v1 (#2275)#22751176 lines
  • chore(roadmap): re-emit the start-command plan sidecar as v2 with its acceptanc…#227615 lines
  • feat(agent-ops): Phase 4 quality readout, writer-model variant and benchmark (#…#22664656 lines
  • test(roadmap): let committed v2 sidecars pass the envelope characterization (#2…#226849 lines
  • feat(agent-ops): Phase 4 section editing, revise, generate and regenerate (#227…#22715457 lines
  • docs(plans): filing records the authored Start command (#2267)#2267251 lines
  • docs(plan): recommend a Compound Engineering start command on every open issue …#2257579 lines
  • feat(roadmap): add shadow, approve and the R14 report for the backlog sorter (#…#22592127 lines
  • fix(drafts): a paraphrase of a single-element record_array no longer raises a c…#2262355 lines
  • docs(audit): record the #2065 record-array matcher follow-ups (#2263)#2263177 lines
  • fix(coherence): a second agreeing reference no longer turns a grounded hedge in…#2260668 lines
  • docs(plans): list entry stable identity plan (#2212 step 2) (#2254)#2254569 lines
  • feat(agent-ops): Phase 4 portal capture, facts, items and sections (#2253)#22535145 lines
  • feat(hooks): the pre-push hook runs only the quick checks; heavy gates stay in …#2256133 lines
  • feat(drafts): state each cited fact at the strength its Fact Kind allows (#2186…#2245940 lines
  • feat(roadmap): filing field-write receipts and their export (#2246)#22461737 lines
  • feat(roadmap): write the Lifecycle flag with authenticated receipts and export …#2252875 lines
  • fix(coherence): a single same-sentence reference about another subject no longe…#2247294 lines
  • feat(hooks): docs-only pushes run only CI's docs path; the hook ledger records …#2244674 linesfolded into the change before it
  • ci: retire the lane-drift alerter (#2243)#22433811 lines
  • feat(worktree-cleanup): owner release of finished worktrees (#2225)#22253449 lines
  • docs(audit): record the #1825 single-reference subject follow-ups (#2248)#2248166 lines
  • test(roadmap): give the 150-issue volume test its own time budget (#2249)#22494 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(ci): adopt the CI change policy: protected gates, structural retirement, n…#224197 lines
  • feat(roadmap): write Horizon, Effort and Priority only from an approved snapsho…#22381750 lines
  • fix(drafts): clear an orphaned dirty-scalar flag in the write that orphans it (…#2229456 lines
  • feat(roadmap): record the intended classification in filed issues (filing align…#2230432 lines
  • feat(agent-ops): Phase 4 inputs — document upload, Drive import, NOFO stage/con…#223715114 linesfolded into the change before it
  • fix(drafts): stop flagging a classifier echo of a number the section already gr…#2240246 lines
  • feat(vault): suggest a Fact Kind for new facts after ingestion, off by default …#2239822 lines
  • docs(strategy): state the generation-quality goal and its ordering (#2189)#2189844 lines
  • fix(drafts): hold a changed list citation instead of re-pointing it (#2226)#2226856 linesfolded into the change before it
  • ci: run only the fast lanes at Ready; the merge queue runs the full pipeline (#…#2228456 lines
  • fix(drafts): keep a length acknowledgment when a rematch leaves the length unch…#2236201 lines
  • test(user-flows): record user-test run 12 and its filed findings (#2194)#2194315 lines
  • fix(autopilot): let reproduce observe admit probes that use Vitest mocks (#2227)#2227464 lines
  • fix(coherence): rounding words only accept a value at or below the claim, and w…#2218283 lines
  • feat(voice): condition section drafts on the organization's own typed edits (#2…#22143134 lines
  • feat(vault): Fact Kind storage, sole writer and classifier core (#2186, PR 1 of…#221714435 lines
  • fix(coherence): keep the warning when a same-sentence reference contradicts a h…#2204401 lines
  • feat(roadmap): filing proposes Priority only for defects and debt (filing align…#2192328 lines
  • feat(roadmap): compile the whole backlog into a hashed snapshot (#2224)#22243495 lines
  • fix(drafts): rematch marks unparseable figures and heals dirty-scalar flags (#2…#2220357 lines
  • test(verify): stop the grandchild-reaping flake in the live-db lane (#2219)#2219139 lines
  • fix(roadmap): stop the Outcome fallback at the end of the Outcome paragraph (#2…#220925 lines
  • feat(agent-ops): add the UI parity inventory, upload intents, and detached runs…#220516603 linesfolded into the change before it
  • fix(regenerate): update a corrected currency/percent figure in place (#2195) (#…#2201521 lines
  • fix(coherence): stop treating a decimal point as a sentence end in hedge ground…#2200155 lines
  • feat(roadmap): project current merged assessments, pins and captured facts; ame…#22031139 lines
  • fix(derived-metrics): ground founding-age and event years on real vault names (…#2202450 lines
  • feat(roadmap): acceptance criteria in filed issues (filing alignment U1+U2) (#2…#21851090 lines
  • fix(cite): compare percent hedges on the coherence scale (#2198) (#2199)#219987 lines
  • test(user-flows): the stalled progress rail was a hidden test tab, not a bug (#…#2162525 lines
  • docs(plans): filing alignment contract plan (#2181)#2181455 lines
  • fix(drafts): keep cited currency/percent figures formatted after vault edits (#…#2191428 lines
  • fix(autopilot): close false-VERIFIED routes in the reproduce gate (R14, option …#21772601 linesfolded into the change before it
  • feat(roadmap): run both seats concurrently, record usage, stop on quota (#2188)#21881308 lines
  • fix(generation): stop writing "[NOTE: not verified]" into section prose (#2190)#219057 lines
  • fix(cite): stop flagging identifier digits and number fragments as unsourced cl…#2184687 lines
  • fix(worktree-cleanup): read growing transcripts by prefix; operator release of …#2183794 lines
  • feat(roadmap): gather Phase B code-scout evidence only where KTD4 needs it (#21…#21801122 lines
  • feat(roadmap): seats return the backlog-ranking contract, merged by one table (…#2179722 lines
  • fix(roadmap): a panel seat's long reason no longer discards its verdict (#2172)#217253 lines
  • feat(roadmap): bind every sorter evidence packet to the Milestone 1 spec (#2171)#2171426 lines
  • fix(agent-ops): judge.run rescores stale scores instead of refusing them (#2170)#217088 lines
  • fix(roadmap): stop the sorter reading a model's "can't tell" as clear (#2169)#21691005 lines
  • feat(agent-ops): flag actions, snapshots, judges and chat over the CLI (#2151)#215139979 lines
  • docs(audit): re-triage the calibration pool under mock policy v1 (U7) (#2168)#2168174 lines
  • test(autopilot): mock-policy attack-pass decision table (U6) (#2161)#21611187 lines
  • ci(deploy): re-sync the production Inngest app after each deploy (#2157)#2157290 lines
  • docs(audits): record the funder-qualification follow-ups and their outcome cand…#2156427 lines
  • feat(autopilot): dependence witness and boundary-doubled verdict; admit mock po…#21552955 lines
  • feat(worktree-cleanup): observe a reach-based doubt rule beside the deciding on…#21541260 lines
  • feat(autopilot): record, approve and recheck the reproduce mock set (U3) (#2153)#21531092 lines
  • fix(worktree-cleanup): judge transcript liveness by each record's own time (#21…#2152361 lines
  • fix: let a generation run finish when its checks cannot run (#2148)#2148368 lines
  • feat(autopilot): vi.mock admission analysis under reproduce mock policy v1 (tes…#21501768 lines
  • feat(autopilot): mutation-site and mutant generator for the dependence witness …#21491014 lines
  • fix(autopilot): refuse every visible route from a reproduce check to vi.mock (#…#21371302 lines
  • fix(agent-ops): grant.intake retries a failed planning run (#2146)#2146483 lines
  • fix(hooks): run one pre-push verify at a time across worktrees (#2144)#21441149 lines
  • fix(roadmap): stop a slow launcher exit from discarding a completed Astra answe…#213655 lines
  • fix(planning): stop asking Claude 5.5 for chain-of-thought output (#2134)#2134152 lines
  • fix(roadmap): deliver sorter notifications when the notifier setting is unreada…#213516 lines
  • feat(roadmap): make Astra and GLM the sorter's seat pair (U6-U8) (#2133)#213333452 lines
  • feat(autopilot): add a non-voting mergeability report beside every verdict (#21…#2126708 lines
  • fix(worktree-cleanup): keep one worktree's hold from ending the night, and say …#21251691 lines
  • feat(autopilot): flag a check's own type errors at fidelity, before freeze (#21…#2124566 lines
  • fix(feedback): refuse GitHub issue drafts for weak-fingerprint reports (#2120)#212056 lines
  • feat(autopilot): add a bounded, pinned typecheck runner for harness worktrees (…#2121481 lines
  • feat(autopilot): name failing tests and label how they failed on every refusal …#2122657 lines
  • docs(audits): execute-and-observe feasibility spike for the check mock policy (…#2123359 lines
  • docs(audits): reproduce calibration reclassification and escape log (#2119)#2119106 lines
  • test: give the live-tree substrate-audit cases a realistic timeout (#2118)#21184 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(portal-paste): call fetch unbound so pasting portal questions works in brow…#211774 lines
  • docs(discovery): disclosure and funder AI-policy follow-ups from the positionin…#2113221 lines
  • fix(roadmap): record an agreed current-value sorter result as no-change (#2112)#2112196 lines
  • docs(plans): reproduce calibration hardening plan (#2111)#2111323 lines
  • fix(agent-ops): close the CLI gaps the production smoke check found (#2110)#2110344 lines
  • feat(marketing): answer the funder-AI fear and name the fit check on the home p…#2109121 linesfolded into the change before it
  • docs: DIY audience research, verified deep research, and positioning working th…#2107483 lines
  • feat(agent-ops): phase 2 — create, intake, generate, and job.status (#2105)#210518638 lines
  • fix(worktree-cleanup): hold chained node_modules junctions at sorting (#2108)#21081166 lines
  • fix(autopilot): give harness test runs a generous timeout and fold the probe ve…#2106113 lines
  • fix(autopilot): stop production modules in a check's closure refusing its freez…#2104326 lines
  • fix(autopilot): stop test titles tripping the red-run infrastructure veto (#210…#2103113 lines
  • fix(pinned-tokens): wait for the writer's commit before extracting (#2100)#2100585 lines
  • feat(agent-ops): token-authenticated agent operations and CLI, phase 1 (#2098)#209829421 lines
  • fix(autopilot): let reproduce observe a pure-function bug beside live-DB tests …#2101288 lines
  • feat(roadmap): judge sorter seats on their decision, not their wording (#2091)#20911084 lines
  • fix(qualification): show the assistant's card, open the project, and explain re…#2097721 lines
  • chore(roadmap): emit the intake-fix leftover outcome candidates (#2094)#2094167 lines
  • docs(ci): record merge-queue activation receipts (#2096)#20968 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(ci): validate queued merges against current main (#2095)#2095318 lines
  • fix(funder-lookup): restore funder qualification facts in production (#2090)#2090145 lines
  • fix(agent): let the intake copilot save the Grant Brief answers it acknowledges…#2088176 lines
  • content(resources): pillar page, annotated sample, demand-based retargeting, an…#2078651 lines
  • fix(intake): let a user fill every field the intake gate requires (#2087)#2087521 lines
  • fix(roadmap): let a merged rolling PR stay history so publishing can resume (#2…#2085120 lines
  • feat(qualification): tell applicants whether and how they can apply before they…#208371242 lines
  • fix(roadmap): launch the ACL PowerShell by absolute path so publishing works (#…#207967 lines
  • fix(roadmap): let the weekly evaluator finish without a human (#2075)#2075517 lines
  • feat(autopilot): add the easy-issue reproduce role (Increment 2) (#2072)#20728227 lines
  • fix(acquisition): make sitemap parsing linear and decode entities once (#2073)#207333 lines
  • docs(audits): record that the production build type-checks the whole repo (#207…#207096 lines
  • docs(audits): writer-model switch follow-ups and their outcome sidecar (#2066)#2066270 lines
  • refactor(inngest): retire the six polling crons and guard the scheduled budget …#20682197 lines
  • fix(deploy): give the Railway build the same 6144MB heap as the CI build lane (…#206718 lines
  • feat(marketing): rebuild the home page for the Mission Steward (#2044)#20442491 lines
  • fix(intake): show the program's name while its action still reaches the agent (…#2060391 lines
  • feat(generation): tune the section prompt for Opus 5.5 and stop flagging brief-…#2061470 lines
  • fix(autopilot): stop a trailing sub-heading becoming extra acceptance criteria …#2059180 lines
  • feat(roadmap): a Claude+GLM panel labels automatic-append cases (#2058)#20582680 lines
  • docs(personas): preserve Grantsville motivations and design guidance (#2053)#20531872 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Rescue worktree documents 2026-09-29T07:08:55.326Z (#2057)#20570 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(llm): switch the writer to Opus 5.5 and main chat to Sonnet 5.5 (#2054)#2054263 lines
  • docs(plans): intake chat transport driver for #1970 (#1974)#1974470 lines
  • docs(audit): funder-lookup EIN-pinned resolution findings (#1984)#1984341 lines
  • docs(content): add the Spiral drafting loop as section 14e (#1971)#1971117 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(analytics): stop a telemetry fault from swallowing the user's click (#1986)#1986164 lines
  • content(resources): add the foundation tax filing guide as a draft (#1973)#1973204 lines
  • Rescued worktree documents (#2039)#20391645 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(llm): move non-writer Sonnet call sites to Sonnet 5.5 (#2051)#205130 lines
  • feat: gate writer-model upgrades behind a real-project comparison (#2047)#20479653 lines
  • feat(roadmap): let the sorter survive stalled seats and change its limits in pl…#20492261 lines
  • docs(audits): provability measurement — 0 of 25 open issues are provable (#1972)#19721407 lines
  • fix(scripts): clean rescue restores under autocrlf and give worktree removal a …#204854 lines
  • fix(ci): give the serial-risk test lanes the heap the ratchet gate needs (#2050)#205058 lines
  • fix(scripts): make the worktree cleanup job finish, report, and not orphan its …#20461493 lines
  • fix(roadmap): stop the sorter rejecting answers its own prompt asked for (#2043)#2043108 lines
  • refactor(inngest): add the stuck-job reconciler and safety-net maintenance in o…#20425352 lines
  • docs(discovery): the policy-change label audit is red on current history (#2032)#2032108 lines
  • feat(pricing): stage prices at checkout and the first-created free application …#20401676 lines
  • fix(roadmap): collect relevant pinned evidence for existing issues (#2041)#2041261 lines
  • feat(scripts): automated daily worktree cleanup with doc rescue and local archi…#20367801 lines
  • fix(intake): make interview chips fit the question (#2035)#20351548 lines
  • fix(roadmap): diagnose assessments and accept equivalent support (#2034)#2034353 lines
  • chore(env): load CHIP_DISPATCH_SIGNING_SECRET in local dev (#2033)#20339 lines
  • fix(intake): count each interview turn exactly once (#2030)#20301429 lines
  • fix(roadmap): let bounded model assessments reach validation (#2026)#2026154 lines
  • docs: skipped required checks count as success; pre-push hook wall-time audit (…#2029242 lines
  • perf(inngest): run the length-budget integrity scan daily (#2025)#2025573 lines
  • fix(ci): classify .agents/ as docs so all-docs PRs skip the heavy lanes (#2020)#202016 lines
  • feat(roadmap): 15 filings per day; automatic close needs 50 labeled cases (#202…#2021126 lines
  • fix(roadmap): read native fields during sorter adoption (#2022)#202226 lines
  • feat(roadmap): apply agreed priorities automatically (#1990)#19906847 lines
  • chore(roadmap): emit the adaptive intake interview follow-up candidates (#2019)#2019196 lines
  • feat(intake): ask only what this grant changes about the selected program (#200…#20017212 lines
  • docs(marketing): home page user test and positioning research, product-marketin…#20082674 lines
  • fix(ci): stop running CI lanes on draft pushes; keep a zero-lane draft guard (#…#20072328 lines
  • feat(hooks): make the pre-push hook the blocking early check (#1999, PR A) (#20…#20043565 lines
  • docs(marketing): Reddit grant-writer corpus crossref, product-marketing v7 (#20…#2005172 lines
  • feat(roadmap): file a session's issues with one command (#1998)#19986101 lines
  • feat(planning): document length budget: one writer ceiling, plan-weighted secti…#19684406 lines
  • fix(ci): stop test lanes re-running files their gate job already runs (#2003)#2003510 lines
  • fix(ci): enforce real-tree lint assertions in the gate CLIs, not only the test …#2002783 lines
  • fix(ci): shard the test lanes so main canaries stop timing out (#2000)#20001394 lines
  • fix(ci): teach deploy qualification and CI monitors the shard-v4 topology (#199…#1995885 lines
  • docs(ci): measure the draft-PR CI tier against what it catches (#1996)#1996113 lines
  • fix(roadmap): unblock shadow validation of CI sentinel examples (#1989)#198967 lines
  • fix(roadmap): tell close approvers satisfied precision is not estimable (#1987)#19872666 lines
  • fix(roadmap): resume failed shadow collector runs safely (#1985)#1985372 lines
  • test(intake): chat transport driver and retry characterization for #1970 (#1980)#19801169 lines
  • fix(roadmap): let completed shadow assessments reach collector completion (#197…#197558 lines
  • fix(roadmap): release the valid mixed existence verdict behind a qualification …#19672801 lines
  • feat(planning): section planning becomes the writer's argument (#1958)#195819443 lines
  • feat(planning): section ledger matches the design mock (#1882)#18822482 lines
  • feat(roadmap): surface each open decision in an issue slice to the right decide…#195912974 lines
  • fix(coherence): stop the entailment judge flagging omission and reformatting as…#1960528 lines
  • docs(discovery): an emptied outcome sidecar records no reason for being empty (…#1951154 lines
  • docs(discovery): land the PR #1838 harness follow-up sources and their sidecars…#1952527 lines
  • fix(roadmap): restore four candidates my R2 refresh dropped without judging the…#1948258 lines
  • feat(review-router): report-only review sorting and status for every PR (#1949)#19491733 lines
  • feat(roadmap): forward the create-budget override through the apply wrapper (#1…#193711 lines
  • chore(roadmap): refresh the drifted sidecars R1 found still open, drop what is …#1938992 lines
  • docs(plans): sidecar evidence refresh — clear the drifted backlog without filin…#1928168 linesfolded into the change before it
  • chore(roadmap): commit the run-11 user-test audits and their sidecars (#1927)#1927530 lines
  • chore(roadmap): commit the sidecars that were sitting uncommitted in the owner'…#18863395 lines
  • feat(roadmap): IF4 — solo operation: the immediate path works end to end for on…#18811517 lines
  • feat(planning): atom ledger with per-fact roles and selection provenance (#1880)#18805885 lines
  • feat(roadmap): IF3 — a damage budget on issue creation, derived from the journa…#18791615 lines
  • feat(roadmap): IF2 — deliver findings from retained commits to a durable inbox …#18771589 lines
  • docs(ces): two-family relevance pass, the generation defect, and the legacy-not…#18691298 lines
  • docs(solutions): compound the hook-dispatch failure into a learning (#1875)#1875126 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): immediate filing — plan, admission contract (U0), single-finding…#18741098 lines
  • feat(planning): section panel guidance with a unified copilot feed and two-phas…#186013695 lines
  • docs(plans): producer emission timing — candidates a scheduled drain can admit …#1873432 lines
  • fix(hooks): register the hooks in the form Claude Code actually dispatches (#18…#186899 lines
  • fix(planning): send Generate to the signing card when the attestation is unsign…#18711078 lines
  • fix(roadmap): the drain captures the existence record its shadow call requires …#1870305 lines
  • feat(roadmap): verify-legacy proves and promotes the imported rows (U2.E) (#186…#18671408 lines
  • feat(roadmap): apply reconciles its own unresolved creates (U2.D) (#1864)#18641301 lines
  • docs(compound): insert-id sanitizer learning + #1453 deferred outcomes (#1863)#1863903 lines
  • feat(vault): seed official documents at onboarding and gate stale filings at ce…#186547450 lines
  • feat(rules): Phase 3 — PreToolUse hooks, heartbeat gate, and the policy-path de…#18662265 lines
  • feat(nofo): let writers correct extracted requirements from the settings board …#18589119 lines
  • fix(ci): give smoke room for runner variance (#1862)#186217 lines
  • feat(roadmap): run-evidence artifact — drain writes it, apply requires it (U2.C…#18591445 lines
  • feat(rules): ratchets, sole-writer gates, and a schema registry promote five le…#18616746 lines
  • feat(roadmap): census ownership proof and working reconcile for intake (U2.A+B)…#185731683 lines
  • feat(claude-md): truth ledger for root rules, rules CI lane, and policy gates (…#18567178 lines
  • feat(roadmap): one pure execution summary for status and apply (U5.E) (#1855)#1855617 lines
  • fix: preserve integrity scan evidence across retries (#1850)#1850432 lines
  • fix: refuse unsafe proposal references before export (#1848)#18481214 lines
  • feat(roadmap): operator resolver and terminal abandoned state for intake (U5.D)…#18541581 lines
  • fix: keep audit summary failures free of tenant details (#1849)#1849533 lines
  • fix: match revision submit shortcuts to the user's platform (#1847)#1847555 lines
  • refactor(ci): simplify CI while preserving validation coverage (#1852)#185220753 lines
  • feat(roadmap): plan-scoped recovery and failure isolation for intake apply (U5.…#18461507 lines
  • docs(audit): computation records have no terminal-write guarantee (#1759)#1759246 lines
  • feat(drafts): narrative card — deliver the org's narrative atoms whole to the d…#18444515 lines
  • feat(roadmap): exclusive execution ownership for the intake journal (U5.B) (#18…#18452387 lines
  • feat(episode): route blocker episodes by cell and repair the cited sentence (#1…#18436047 lines
  • docs(solutions): pair conditioning with writing capability, per step (#1842)#1842317 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Fix #1535: verified candidate fifth-real-1535-20260914 (#1841)#1841394 lines
  • feat(roadmap): one shared execution journal per repository (U5.A) (#1839)#18391890 lines
  • fix(roadmap): scheduled collector writes the per-entry existence record it requ…#1840207 lines
  • feat(autopilot): evidence harness — Increment 1 (#1838)#18388200 lines
  • feat(voice): arm substance selection for all orgs, and correct the comment that…#1771347 lines
  • docs(plans): land the #1824 phase-gate plan that PR #1835 references (#1837)#1837288 lines
  • feat(roadmap): durable execution completion for intake apply (U1) (#1834)#18343312 lines
  • fix(agent): pair house-style conditioning with writing capability, per step (#1…#18351608 lines
  • feat(coverage): cited evidence spans — the writer reads their own sentences (#1…#18336802 lines
  • fix(ci): run production build alongside static checks (#1836)#18361601 lines
  • docs(brainstorm): land the blocker-guidance research so its findings have a hom…#17721383 lines
  • docs(plans): say where the system looked, instead of saying nothing (#1830)#1830167 lines
  • feat(voice): give the org's confirmed specification first claim on the prompt (…#1831952 lines
  • fix(voice): make the preference budget count what it emits (#1670) (#1775)#1775482 lines
  • docs(plans): cited evidence spans, and the gate decisions it unblocks (#1774)#17741015 lines
  • chore(workspace): remove the orphaned metric-bodies directory (#1782)#17822345 lines
  • feat(roadmap): implement the intake approval classification transport (#1776)#17761192 lines
  • fix(roadmap): verify existence at execute time instead of pinning a SHA (#1773)#17731699 lines
  • feat(workspace): returning to a requirement resumes it, rather than restarting …#17691095 lines
  • feat(roadmap): name a candidate's filed lineage on the draft, deterministically…#1770880 lines
  • fix(vault): label product-seeded compliance atoms as system provenance (#1767)#17679667 lines
  • feat(voice): web-verbatim boilerplate census against Common Crawl, and the 09-1…#17661490 lines
  • feat(workspace): retire an Episode that outlives its blocker (U6) (#1765)#1765359 lines
  • feat(workspace): open a blocker Episode with visible scope (U5) (#1762)#17621698 lines
  • fix(roadmap): drain recovery that cannot wedge on the surface that is failing (…#17542728 lines
  • feat(roadmap): a model verdict annotates a candidate; only marker identity may …#17641413 lines
  • feat(roadmap): bounded evidence view makes the assessment payload scale with th…#17611501 lines
  • feat(chip-dispatch): blocker token purpose and redemption authority (U4) (#1760)#17601623 lines
  • feat(roadmap): assessment on an owned subscription seat, and no silent suppress…#17581226 lines
  • feat: Final Checks pre-certification stage (19 of 21 units, three review rounds…#17428529 lines
  • feat(review-chat): blocker identity, envelope, and working scope (U3) (#1757)#1757841 lines
  • feat(agent): one outstanding count across the copilot and the checks panel (U8)…#1755428 lines
  • fix(ci): give the DB-free lanes their sentinel locally, not just in CI (#1756)#1756249 lines
  • docs(roadmap): the assessment boundary was real, and #1734 dissolved it properl…#173779 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(roadmap): count recovered predecessors against the daily send budget (#1753)#1753213 lines
  • feat(workspace): render the check breakdown in the copilot feed (U2) (#1751)#17511021 lines
  • docs(plan): defer U4, rescope U5, and correct two stale findings (#1752)#175259 lines
  • feat(roadmap): versioned atomic retry ledger (U3) (#1749)#17491375 lines
  • feat(workspace): publish a per-check breakdown from the gate projection (U1) (#…#17481123 lines
  • feat(roadmap): prove host ownership with an OS-arbitrated endpoint (U2) (#1747)#1747442 lines
  • feat(roadmap): scope marker incidents to their candidate (U1) (#1745)#1745674 lines
  • docs(plan): intake recovery and marker semantics — five units (#1744)#1744554 lines
  • docs(plan): supersede the recovery ladder, hold U6, land the review evidence (#…#1743796 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): cross-family duplicate sweep gate (U1) (#1740)#174011974 lines
  • docs(plans): #1673's repair tranche is measured out of existence (#1722)#1722411 lines
  • docs(plan): correct the Phase 2 gate — do not widen the eligibility signing win…#1739237 lines
  • feat(workspace): Final Checks owns non-draft conditions (Phase 1 — blocker rout…#17382552 lines
  • fix(roadmap): refuse a calibration cohort whose zero-error clauses are vacuous …#1736180 lines
  • fix(workspace): an outline row names exactly what the assistant edits (#1735)#17357632 lines
  • docs(audits): the voice census measured an org that was never in the cohort (#1…#1733496 lines
  • feat(roadmap): fill the intake assessment with the pinned model, or refuse (#17…#17341248 lines
  • docs(plans): record the smoke-lane arc's outcome and file its remaining work (#…#1732445 lines
  • docs(discovery): capture three CI-alerter findings deferred from #1728 (#1731)#1731463 lines
  • docs(discovery): record what the docs-gates fix taught, and exercise the lane (…#172945 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(ci): stop docs-gates from blinding the drift alerter (hotfix, explicitly sc…#172880 lines
  • test(ci): give the stale-plan auditor its missing fixtures, and ratchet checker…#1730626 lines
  • feat(seo): cross-link the resources cluster, add nav Guides item and article br…#1726159 lines
  • feat(ci): widen the dormancy signal to every unconsumed callable (#1727)#1727375 lines
  • feat(ci): ratchet the tested frontier, and take both dormancy scans off the smo…#1725798 lines
  • feat(resources): link the hub from the footer and add two guides (#1720)#1720321 lines
  • feat(ci): assert registered mechanisms have a production consumer (#1723)#1723989 lines
  • docs(discovery): a docs-only PR skips the lint that governs docs (#1719)#1719237 lines
  • fix(ci): make the gate that governs producer docs reachable, without slowing do…#1721328 lines
  • feat(roadmap): existence-verification machinery for intake — detector measured …#17179140 lines
  • docs(solutions): capture the two durable learnings from the U7 census (#1714)#1714227 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(workspace): Proposal Workspace shell — spine, projection-backed checks (#1…#169253170 lines
  • feat(resources): publish the funder-access guide (#1716)#1716182 lines
  • fix(ci): give the lane-drift alerter hysteresis so it stops re-alerting and can…#1718223 lines
  • perf(ci): give the smoke lane real headroom (U1-U4) (#1715)#17152792 lines
  • docs(solutions): compound two learnings from the silent-main-red incident (#169…#1690255 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(brainstorm): two-path issue-creation reconciliation finding (#1708)#1708421 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): give intake holds a reader so filed work cannot be silently lost…#1711480 lines
  • feat(security): rebuild the server-action census on ts.Program + TypeChecker (#…#17063471 lines
  • fix(roadmap): quarantine stale evidence at intake drain preparation (#1712)#1712553 lines
  • docs(solutions): compound two learnings from the ACL creation gap (#1700)#1700147 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(roadmap): stop truncating four-digit plan ids in the stale-plan audit (#170…#170911 lines
  • feat(roadmap): scope the intake write credential to one child invocation (#1707)#170793 lines
  • feat(roadmap): render a fileable Inbox draft for a held candidate (#1705)#1705349 lines
  • docs(roadmap): record the tenure-detection plan and its six outcome candidates …#16931285 lines
  • feat(roadmap): GitHub-backed operator intake approval transport (#1704)#1704359 lines
  • voice: composite census, prompt de-identification, and intake candidates (#1701)#17012182 lines
  • fix(roadmap): make per-candidate drain dispositions survive collection (#1699)#1699124 lines
  • feat(security): flag bare organization reads in API routes (#1240 U7, part 1) (…#1695293 lines
  • fix(roadmap): carry issue bodies into the intake snapshot projection (#1698)#169831 lines
  • fix(roadmap): three defects found by the first live drain run (#1696)#1696458 lines
  • feat(roadmap): operator intake approval, apply, and one-drain-one-plan (#1689)#16891965 lines
  • fix(ci): correct the canary watchdog after two review rounds (#1688)#1688894 lines
  • feat(roadmap): operator-initiated intake reaches an assessment (#1678) (#1687)#16872764 lines
  • fix(ci): close the 2026-08-31 silent-main-red incident — flaky guard + canary w…#16861197 lines
  • test(auth): pin the onboarding membership gate; make the refusal page honest (#…#1683411 lines
  • fix(roadmap): close the Windows private-directory ACL creation gap (#1684)#16841686 lines
  • fix(auth): revalidate membership roles at request time (#1682)#16821102 lines
  • feat(analytics): route PostHog exceptions into canonical roadmap intake (#1681)#16811460 lines
  • fix(roadmap): make the collector's existing per-entry resume reachable (#1679)#1679146 lines
  • fix(coherence): detect organizational tenure claims by attachment, not vocabula…#16802755 lines
  • docs(content): document the operating loops and scope the two SEO ledgers (#166…#1666132 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(roadmap): plan the collector resume that already exists but cannot be reac…#1667310 lines
  • feat(resources): publish the NOFO guide after source verification (#1665)#1665161 lines
  • fix(roadmap): resume the trusted initial inventory instead of colliding on it (…#1664109 lines
  • docs(solutions): unconsumed once-stub leaks across tests under clearAllMocks (#…#1659131 linesfolded into the change before it
  • docs(solutions): record that a roadmap contentSha256 is not one thing (#1661)#166175 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(marketing): add the /resources content route (#1663)#16633181 lines
  • fix(roadmap): make weekly reconciliation runnable and collector failures diagno…#1662630 lines
  • feat(roadmap): name the judge, define routine eligibility, and let judgements b…#1660587 lines
  • feat(roadmap): preserve the competing issues a hold was made against (#1658)#1658239 lines
  • fix(coherence): ground user-typed brief figures and hedged citations (#1657)#1657860 lines
  • feat(roadmap): capture judgements against banked shadow observations (#1656)#16561580 lines
  • feat(coherence): all-orgs cohort scope for evidence suppression; commit placeho…#1653237 lines
  • fix(seo): real 404 for unknown paths and Organization logo markup (#1654)#16541716 lines
  • fix(coherence): bind numeric claims to evidence rows; suppression behind off/sh…#165216719 lines
  • fix(roadmap): say "not measured" instead of zero, and count every disagreement …#1651128 lines
  • Roadmap shadow observations <redacted>…#1644107327 lines
  • docs(roadmap): record first retrospective cohort human adjudication (#1647)#164736511 lines
  • feat(roadmap): name verification recency and origin/main binding separately (#1…#1650563 lines
  • feat(roadmap): make classifier accuracy measurable and enforce the enable-flag …#16481608 lines
  • docs(user-test): record run 3 — flag quality and the new-grant blocker (#1643)#164374 lines
  • feat(roadmap): pair historical and live calibration cohorts (#1646)#16461460 lines
  • fix(agent): accept Better Auth organization IDs (#1645)#164511 lines
  • fix(roadmap): pin publisher runtime digest (#1642)#164242 lines
  • fix(review): hide internal diagnostics from flag tooltips (#1638)#1638657 lines
  • fix(roadmap): separate publisher authority (#1637)#1637731 lines
  • fix(review): keep synthetic spans out of inline text (#1636)#1636163 lines
  • fix(roadmap): bind activation evaluator runs (#1635)#163526 lines
  • fix(roadmap): secure promoted Windows spool entries (#1634)#163431 lines
  • fix(roadmap): allow archived assessment schemas (#1633)#16332 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(user-test): record run 2 and the run-2b flag-actions pass (#1619)#1619115 lines
  • fix(review): repair disputed facts with family-aware routing (#1630)#16302333 lines
  • fix(roadmap): complete shadow collector handoff (#1632)#163246 lines
  • fix(roadmap): permit bounded evaluator file reads (#1631)#163139 lines
  • fix(roadmap): enforce collector terminal reserve (#1629)#1629325 lines
  • fix(review): restore actionable flag resolution (#1624)#1624515 lines
  • fix(deploy): enforce live runtime configuration readiness (#1628)#16281951 lines
  • fix(roadmap): align collector weekly identity (#1627)#162747 lines
  • fix(roadmap): seal snapshot file bytes (#1626)#162620 lines
  • fix(roadmap): seal candidate file bytes (#1625)#162514 lines
  • fix(roadmap): bind candidate to production repository (#1622)#162220 lines
  • fix(roadmap): support fine-grained collector reads (#1621)#162199 lines
  • fix(roadmap): require issue-fields read authority (#1616)#161629 lines
  • fix(roadmap): isolate scheduled collector runtime (#1615)#1615236 lines
  • fix(roadmap): bind scheduled token before GitHub access (#1614)#1614239 lines
  • fix(roadmap): run collector from disposable checkout (#1613)#161316 lines
  • feat(roadmap): add hands-off two-stage shadow evaluation (#1612)#16123183 lines
  • fix: normalize requested funding across grant workflows (#1611)#16112871 lines
  • fix(roadmap): harden Windows shadow profile (#1610)#1610248 lines
  • fix: keep foundation LOIs recoverable through intake (#1609)#1609959 lines
  • feat(roadmap): preserve shadow adjudication operands (U7) (#1608)#16081054 lines
  • fix(review): align Copilot with reader-visible draft text (#1607)#1607315 lines
  • feat(roadmap): automate shadow evidence collection safely (#1600)#160011521 lines
  • fix(auth): enforce deletion admission across product surfaces (#1601)#16019561 lines
  • feat(drafts): disputed facts join the normal flagging system (#1448 shape A) (#…#15974860 lines
  • feat(roadmap): bank shadow calibration observations (#1596)#15961211 lines
  • docs(roadmap): stage 1b calibration evidence — narrowed to the observation half…#1594450 lines
  • docs(solutions): --custom writes a pass-through snapshot, never for schema chan…#1331146 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(roadmap): arm the producer cutover now that shadow drafting has landed (#…#159253 lines
  • feat(roadmap): recommendation-only shadow mode for issue intake (#1591)#15916648 lines
  • fix(roadmap): score automation coverage where automation was possible (#1590)#159083 lines
  • docs(solutions): two learnings from the U8 lane-drift arc (#1589)#1589249 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(ci): give each positive path filter its own key — U8's carve-out never work…#1588235 lines
  • fix(roadmap): let roadmap:candidate accept a producer draft, and refuse a forge…#1587102 lines
  • feat(roadmap): prove the producer path end to end on one real artifact (#1586)#1586156 lines
  • docs(ci): reconcile the runbook to the post-split end state (plan U11) (#1585)#158587 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): a local lane comparison without the sentinel is not a comparis…#158456 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(auth): request-time membership gate — API seam, page seam, and rollout con…#15821708 lines
  • docs(roadmap): correct the post-merge plan status and capture two review learni…#1583539 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): capture five learnings from the worktree/branch cleanup arc (#…#1577408 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): AI-managed roadmap intake (U0-U8), production-disabled (#1581)#158110862 lines
  • fix(scoring): make project scores project-scoped and atomic (#1580)#15805620 lines
  • fix(roadmap): share canonical capability ordering (U1) (#1578)#157879 lines
  • docs: land rescued plans, solutions and user-flow records (#1576)#15764737 lines
  • docs(solutions): a shipped mitigation is not the fix — don't auto-close the iss…#1466162 linesfolded into the change before it
  • docs(plan): correct the #1448 claims and record what it still owns (#1575)#157510 linesfolded into the change before it
  • docs(solutions): a stated guarantee is a test manifest (#1573)#1573125 linesfolded into the change before it
  • docs(solutions): the pin-one-level-off pattern, and a stop rule that can be for…#156587 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(drafts): render the gap-facing altitude offer (R30) (#1570)#15702259 lines
  • fix(ci): scheduled lane-drift alerter (plan U8) (#1572)#15723071 lines
  • feat(roadmap): add deterministic Steward v2 review gate (#1571)#157119062 lines
  • fix(auth): membership accessor + narrow document-route auth fallbacks (SEC-AUTH…#15692020 lines
  • docs(plan): record website-intake completion state; capture the deferral patter…#1554218 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(roadmap): voice roadmap issue-creation proposal, revision 2 (#1566)#1566276 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(audits): commit the July system-state and August coherence audits (#1568)#1568743 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan): record that R30's surface is deferred to #1448 (#1555)#15553 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(ci): collapse the duplicate security-suite execution (plan U7) (#1553)#1553497 lines
  • docs(u12): correct stale issue ownership, record the altitude-pair hazard (#154…#154077 lines
  • docs(roadmap): publish operational cutover evidence (#1551)#15519806 lines
  • fix(workspace): normalize encoded section identities before review-chat chip sc…#1416269 lines
  • docs(solutions): a cast on a db-test fixture insert destroys the validation the…#1510191 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(ci): alert on deploy failure and clear it on green (U9) (#1515)#15151007 lines
  • feat(drafts): gap-facing altitude prompt (U12) (#1519)#15191906 lines
  • Reconcile intentional Inbox issue backlog (#1539)#153936417 lines
  • fix(scoring): enforce authoritative LOI length units (#1429)#14291362 lines
  • feat(voice-bench): stage-one blind A/B voice harness — pair generation, concier…#13415362 lines
  • docs: reconcile manual-sensitive issue roadmap (#1506)#1506977 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(scripts): add the quiescence preflight the repair has been missing (U5 step…#1503540 lines
  • test(inngest): live-DB proof that the projects-row lock serializes a funder reb…#1500184 lines
  • test: assertions that bind — parity contract, telemetry, renders, runtime guard…#14991634 lines
  • feat(ci): split the db lane into three concurrent lanes (U5) (#1491)#14914208 lines
  • fix(drafts): run the seed envelope-unit check on house_default too (U14) (#1497)#1497384 lines
  • docs(roadmap): classify full open issue census (#1498)#149812743 lines
  • docs(roadmap): record second routine steward batch (#1495)#1495463 lines
  • feat(corrections): emit a real value delta so repaired facts refresh their pros…#1496225 lines
  • feat(drafts): carry atom_id on atom_refs and target regeneration by it (#1461) …#1494282 lines
  • docs(roadmap): record first routine steward batch (#1493)#1493697 lines
  • docs(roadmap): record production steward batches (#1492)#1492597 lines
  • feat(roadmap): enable guarded production issue writes (#1490)#1490471 lines
  • fix(ci): enforce package gate classification closure (#1457)#1457186 lines
  • docs: read mergedAt before autoMergeRequest when checking arming state (#1486)#14862 linesfolded into the change before it
  • docs(ci): sample 5 — the rise was noise; the lesson is the sample count (#1485)#148518 linesfolded into the change before it
  • docs: never git checkout -- over uncommitted work in a cleanup loop (#1484)#14841 linesfolded into the change before it
  • docs(ci): the count is 76, and the rule needed the gap closed (#1489)#148922 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(ci): reconcile KTD7 — the set is 51 files, not 12–20 (#1487)#148765 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): add AI-managed GitHub issue control plane (#1483)#14832283 lines
  • feat(backfill): scope the repair to named items and refuse an unexpected blast …#1479863 lines
  • docs: close U10 with both readbacks, plus three corrections from the U4 session…#148256 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(inngest): scope passage regeneration to the corrected vault row (#1461) (#1…#14681233 lines
  • docs(ci): sample 4 fired the watch rule — and it invalidates U5's DB-free gate …#148157 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(roadmap): add AI-managed issue control plane (#1469)#14695492 lines
  • docs(ci): U4 discharged — three canaries, caps derivable, two assumptions close…#147879 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(backfill,ci): #1431 workstream 1 — repair-script safety and the CAS shape g…#14652600 lines
  • docs: armed-but-red PRs do not count against the one-armed-PR limit (#1476)#14762 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(ci): record U4 canary 1 — shape settled at three lanes, caps still open (#…#147584 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(ci): route the timing report through an injected sink, not the console (#14…#1474128 lines
  • fix(ci): capture per-project Vitest timing with a hard-fail guard (U4) (#1473)#14731723 lines
  • docs(solutions): clustering groups symptoms; only a trace names the cause (#147…#147284 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(ci): correct the U12/U13 mechanism — falsified by trace (#1471)#147185 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(ci): fold U4 partial timing into the lane-topology plan (#1470)#1470116 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • perf: stop preloading the tsconfig program in the shared lint scanner (#1467)#1467461 lines
  • feat(vault): U11 — correction capture and repair (#1433)#14335893 lines
  • fix(generation): advance live progress (#1439)#1439652 lines
  • fix(ci): reject duplicate gate commands (#1455)#1455167 lines
  • fix(intake): keep funder lookup misses recoverable (#1456)#1456343 lines
  • fix(design): render DM Sans by moving its bridge variable to <html> (#1454)#145422 lines
  • fix(workspace): gate certification on score hydration (#1430)#14301140 lines
  • fix(workspace): converge regenerated draft text (#1446)#14462666 lines
  • fix(vault): a total tie is a judgment the strip must show, and cover its reader…#1444603 lines
  • fix(drafts): generation seed carries the length-budget provenance tuple (#1432)#14324304 lines
  • fix(intake): clamp staged NOFO free-text so silent confirm cannot 500-soft-lock…#1415206 lines
  • fix: keep intake readiness polls authoritative (#1438)#1438593 lines
  • ci(gates): host CI_GATES manifest assertions in an existing fast-tier gate (#14…#1440493 lines
  • feat(db): Phase D substrate — disputed_at, flag judge identity, correction lint…#139610503 lines
  • fix(vault): enforce dismissal authorization at server boundaries (#1418)#1418691 lines
  • fix(workspace): replace exact selected passage (#1417)#1417410 lines
  • <redacted>#1413106 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(workspace): keep first editor touch silent (#1408)#1408252 lines
  • test(workspace): cover LOI section rendering (#1392)#139266 lines
  • <redacted>#134353 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(scripts): standing census for the #1323 costume-row shape (#1346)#1346322 lines
  • fix(repair): correct dry-run oracle org ids; record U10 tenant repair execution…#1354454 lines
  • feat(vault): attention strip — the judgment calls, not the facts (U10) (#1394)#13942166 lines
  • fix(workspace): allow limitless metadata saves (#1385)#138516 lines
  • docs(solutions): the cross-org UPDATE test proves the SELECT policy, not WITH C…#1393111 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): altitude audit — per-fact verdicts and page-level flags (U9) (#138…#13895002 lines
  • fix(drafts): match percent-family tuple units (#1384)#1384257 lines
  • fix(certification): close the round-2 review findings on the content witness (#…#13721025 lines
  • docs(solutions): verification that confirms instead of discriminating (#1390)#139062 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • test(vault): guard taxonomy header counts (#1375)#137543 lines
  • test(workspace): isolate GuidedCenter server action (#1365)#13656 lines
  • docs(conventions): enumerate numeric validity guards (#1368)#136826 linesfolded into the change before it
  • docs(ideation): editing-page jigs — read-out before dial (#1353)#1353571 linesfolded into the change before it
  • docs(solutions): land four captured learnings that never reached main (#1387)#1387434 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): surface website provenance — excerpts, page grouping, altitude (U8…#13703442 lines
  • feat(vault): altitude-audit substrate — verdict tables, RLS, gates (Phase A2) (…#137620623 lines
  • fix(certification): suspend Budget Coherence until it has a writer (#1382)#13821350 lines
  • feat(workspace): document-scoped review center column (#1381)#13818986 lines
  • fix(ci): make a timed-out lane report as a failure, not a supersede (#1371)#13711060 lines
  • fix(certification): refuse to certify against superseded scores (#1361)#136110536 lines
  • feat(intake): website acquisition through verified provenance (Phase B) (#1357)#135720017 lines
  • ci(db): raise the db lane cap to 20 minutes, and say plainly that it is not a f…#136718 lines
  • fix(types): derive vault row shapes from schema (#1362)#136232 lines
  • feat(vault): website intake substrate — provenance and labeled-example tables (…#135630231 lines
  • feat(vault): complete the org/program parallel model — capture, registry, repai…#134915234 lines
  • docs(marketing): Product Marketing Context v6 — NonprofitAF crossref edits (#13…#134445 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: operator repair for org facts misfiled under prog.* (U1-U6, #1323) (#1342)#13429653 lines
  • docs(solutions): four verification corollaries from the voice observability wor…#1329155 lines
  • docs(plans): re-namespace org facts filed under prog.* (#1323) (#1339)#1339830 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(strategy): NonprofitAF crossref revision — outcome-claim discipline, quali…#1340227 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): hold U10 — locked atoms make clear-and-re-extract inapplicable (#1…#133812 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix: gated org-profile repair (U10) — operator-decided, no automated replacemen…#13353300 lines
  • feat(voice): substance selection as a product rule, replacing the frozen dictio…#13342024 lines
  • feat(voice): substance-selection arm — org-native gap experiment (U1-U5, U7a) (…#129013212 lines
  • fix: activate org_profile classification + close the reclassification window it…#13304787 lines
  • fix: serialize per-org program resolution with an advisory lock (#1282) (#1327)#13272824 lines
  • feat(voice): loud-fail on a stale judge rubric instead of scoring behind a warn…#1324903 lines
  • docs(audits): record the org-name collision census production baseline (#1326)#132691 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(voice): rendered-state trace, derive-only criteria activation, declined-br…#13172190 lines
  • docs(plans): refresh release-2 units against merged main (#1325)#132522 linesfolded into the change before it
  • docs(solutions): capture three learnings from the org-name guard work (#1318)#1318135 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): capture two learnings from the codex-dispatch work (#1322)#1322143 linesfolded into the change before it
  • docs(seo): Read 1 results — inconclusive (n=18 < 30 floor), Packet 003 crawl st…#1256167 linesfolded into the change before it
  • docs(solutions): a check that cannot fail — second occurrence and two technique…#129798 linesfolded into the change before it
  • docs(solutions): three learnings from the portal-field edit-conflict rewrite (#…#1321726 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(codex): stop silent stalls in Claude-to-Codex dispatch (#1319)#1319781 lines
  • fix(portal-field): scope metadata edit conflicts to the changed field group (#1…#131212067 lines
  • fix: org_profile document type and org-name program guard (release 1) (#1313)#131315101 lines
  • feat(voice): make the preferences renderer honest and evict the global slop lis…#13071556 lines
  • docs: portal-field conflict + matcher scoping plans, program-binding investigat…#1300890 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): park the legacy-tuple retirement plan as research (#1301)#1301433 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(monitor): length-budget integrity scan and guard counters (#1291)#12911802 lines
  • fix: scope portal-matcher and intake atom lookups to the grant's program (#1293)#12932346 lines
  • docs: compound stale-mirror learning + codify merge discipline (#1298)#129862 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(workspace): drop the Required chip from the outline rail (#1289)#1289103 lines
  • fix(deliverables): backfill attachments that claim found but hold no document (…#1287539 lines
  • fix(workspace): deliverable affordances belong in the detail panel, not the out…#1285416 lines
  • docs(voice-bench): run-2 evidence corrections — compression probe primary recor…#128329 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(voice): KTD11 — rollout cohort to dedicated audit-trailed table (#1279) (#…#12819167 lines
  • docs(voice-bench): v2 brainstorm seed — run-2 failure signature packaged for th…#128015 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(voice-bench): december-newsletter de-fuse reconciliation (#1278)#127812 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(voice-bench): run-2 Q1 verdict — conditioned arm fails CK 0-4, EHI inconcl…#127665 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(onboarding): make the first LOI journey measurable (#1268)#12685618 lines
  • docs(voice-bench): EHI coverage manifest provisioning (run-2 prereq) (#1266)#1266367 linesfolded into the change before it
  • fix(voice-bench): make blind packs comparable (#1269)#1269473 lines
  • fix(llm): route every document fence through one escaping helper (#1267)#1267203 lines
  • feat(documents): document-driven program facts — U6, U6.5, U7, U8 (#1236)#123624578 lines
  • docs(voice-bench): U6A corpus harvest — CK financial-voice + newsletter registe…#1230796 lines
  • fix(planning): make the R2a ceiling gate native-unit aware (#1264)#1264935 lines
  • feat(marketing): connect free LOI signup path (#1257)#1257929 lines
  • docs(concepts): requirement class, period-sensitive, auto-link (#1248)#12489 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(marketing): establish Grantsville SEO and AIO foundation (#1249)#12493330 lines
  • feat: inline deliverable document linking (#1242)#12428278 lines
  • docs(concepts): add Prompt Rendering vs Storage Rendering + compound-atom learn…#1238238 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): stop "... [TRUNCATED]" preview markers leaking into persisted cita…#12351153 lines
  • docs(seo): link the deferred middleware-404 observation to #1233 (#1234)#12344 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(seo): 2026-07-28 GSC indexing notification assessed — expected, no code ch…#123267 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: draft length budget system — honest ceilings, targets, and the sculpt tai…#123112133 lines
  • feat(planning): sibling-aware missing-atom repair — signal + repair session (#1…#122914905 lines
  • feat(voice): section-conditioned voice register mechanism (U1–U6B) (#1228)#122830426 lines
  • feat(voice): Real-Org Voice Bench — fixtures, corpora, blind-pack script, pre-r…#12053175 lines
  • fix(drafts): tiered percent-family fallback in selector re-bind display matcher…#12271329 lines
  • feat(analytics): PostHog revamp — honest surface, catalog-as-memory, drift audi…#12213138 lines
  • docs(solutions): planning.started emitter-arming convention (from PR #1212) (#1…#1222130 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(agent): arm the primary LOI outline attempt on copilot INTAKE→PLANNING (#12…#1212206 lines
  • fix(db): route every SQLSTATE read through getPgError; allow repeat EIN registr…#12199463 lines
  • docs(seo): measurement-reads runbook — three scheduled reads, crawl-date detect…#1220117 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(workspace): remove always-on authorship visuals (PR2 — rail retired) (#120…#12071309 lines
  • fix(vault): indexed selector resolution for schema-unlocked record-array atoms …#12068428 lines
  • docs(compound): pipe-masked exit codes, build-verified route grammars, SEO prog…#1218118 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan): indexed selector resolution fix for schema-unlocked record-array at…#1203462 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: land PR #1210's plan artifact and CONCEPTS vocabulary (missed the impleme…#1217439 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(seo): Packet 003 treatment deployed and production-verified (#1216)#121611 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(seo): Packet 003 title/meta treatment — approved category title + grounded…#121475 lines
  • docs(solutions): compound PR #1210 learnings — server-only barrel imports, purp…#1215111 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(copilot): generative workspace panel — episodes, problem projection, whats…#121016841 lines
  • docs(seo): Packet 001 verified — GSC domain property, DNS TXT, sitemap submitte…#121316 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(seo): Deployed rung production-verified — all 8 Packet 002 probes passed a…#121112 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(seo): ledger Merged rung for PR #1208 (#1209)#120910 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(seo): Packet 002 crawl, canonical, and indexation foundations (#1208)#12081433 lines
  • docs(plan): authorship-provenance display plan (PR1 #1202 shipped; rail deferre…#1204178 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(agent): let the agent edit human-edited passages (#1202)#1202198 lines
  • fix(cite): correct cite-missing-attribution over-fire — measure, fail-open, bac…#11996179 lines
  • spike: claim-emission measurement harness + pre-registered verdict (NO-GO as sp…#116131644 linesfolded into the change before it
  • docs(solutions): compound the no-NOFO LOI intake pattern (#1107)#1107323 linesfolded into the change before it
  • docs(maps): refresh system map for the workspace overhaul + add HTML mirror (#1…#1079464 lines
  • test(drafts): repoint NOFO-confirm guards to unified confirm-nofo.ts writer (fi…#1201426 lines
  • chore(reextract): add CK pre-merge diagnostic gate (#955)#95525 lines
  • feat(grants): make application documents user-owned (#1198)#119827788 lines
  • feat: structured section drafts — closed structural vocabulary with WYSIWYG edi…#11926357 lines
  • feat(api): per-user rate limit on /api/ein and manual-org-request (#558) (#1197)#1197416 lines
  • test(workspace): extract DocumentComposer create-flow helpers with behavioral c…#1196683 lines
  • fix(lifecycle-email): two-phase quick-win claim — pre-LLM candidate claim + rec…#11958779 lines
  • feat(vault): require non-opted record_array sub-fields at the Zod boundary (#47…#119494 lines
  • refactor(vault): retire CompoundValue runtime scaffolding the discriminated uni…#1193650 lines
  • fix(drafts): clamp mapRange terminal zero-width mapping to raw EOF (#1174) (#11…#118853 lines
  • test(auth): assert lockfile-resolved better-auth version meets the magic-link p…#118651 lines
  • refactor(drafts): dedup remap-or-fallback degrade contract into remapSpanOrFall…#1185101 lines
  • fix(ci): harden and optimize delivery pipeline (#1191)#11913062 lines
  • docs: recover portal-generation-identity plan artifact + second carry-list occu…#1190321 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: type-less portal-field generation with question-driven atom matcher (#118…#11831689 lines
  • docs(solutions): pipeline-owned section-field checklist from PR #1181 review ar…#118972 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(planning): atom chip semantic contract — inclusion baseline + two-axis usa…#11812699 lines
  • test(drafts): production-shape coverage for offset-map byte-identity guard degr…#1187154 lines
  • ci: bump db-lane timeout 13->15min (vitest grew to ~11m50s) (#1184)#118418 lines
  • feat(planning): section regeneration commit (U1–U6) (#1180)#11802963 lines
  • feat: portal field question-driven generation (#1179)#11792850 lines
  • docs(brainstorm): add provenance-spine decision addendum to claim-ledger origin…#1171171 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): ce-compound-refresh sweep of docs/solutions + CONCEPTS.md (#11…#11771704 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): anchor resolved-space flag spans to raw at both call sites (foundi…#11701564 lines
  • fix(drafts): strip leaked compound-catalog placeholder ref tokens before persis…#1169294 lines
  • fix(vault): compound-atom catalog renders full data instead of example menu (#1…#11681624 lines
  • fix(planning): collapse single-field portal item into one leaf row (#1164)#11641063 lines
  • fix(editor): make batch detach-confirm dialog legible (#1160)#1160106 lines
  • fix(workspace): re-check indicator hides on never-evaluated sections - checking…#1159116 lines
  • fix(drafts): stop rematch writing prefetch prompt guidance into resolved_value …#1158407 lines
  • fix(editor,workspace): bind fetch to globalThis - unbound method calls threw Il…#115740 lines
  • feat: in-place surgical section editing with inline flags and verify-on-touch (…#115617681 lines
  • fix(intake): paste-channel grants reach PLANNING + persist pasted NOFO (#1154)#1154440 lines
  • fix(outline): hide dead "Add section" control on non-appendable documents (#115…#115582 lines
  • feat: folders as submission buckets — parent-pointer membership, gate-relevant …#115223023 lines
  • docs(solutions): source-invariant proxy scoping + llm_call_audit abort-noise tr…#1153139 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(roadmap): Anthropic-safe emit_roadmap tool schema (named-field looseObject,…#1151344 lines
  • feat: intake tightening + lifecycle legibility (#1150)#11503647 lines
  • docs(solutions): useGenerationProgress treats PLANNING as failed generation — d…#114996 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(planning): render structured vault values as JSON in the remediation overla…#114825 lines
  • fix(ui): require dialog-element target for Modal backdrop dismissal (#1147)#114754 lines
  • docs(solutions): transient create action belongs in an overlay, not a nav-node …#114280 linesfolded into the change before it
  • feat(planning): auto-include matched vault atoms at plan time (#1137) (#1146)#11463626 lines
  • feat: Type-first progressive-disclosure document composer with auto-planned gen…#11434004 lines
  • docs(solutions): seeding plan.dataSelection — mechanism choice + re-derive idem…#1145137 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): section-readiness-card v2 plan (split; #3 design reference for #11…#1144488 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(planning): section-card inline chip delete + missing-item remediation over…#11381584 lines
  • feat: Type-aware document creation at the planning stage (#1136)#11362326 lines
  • docs(solutions): shared-component scroll-ownership convention (workspace panels…#113094 linesfolded into the change before it
  • docs(solutions): composer generationMode is create-time-immutable; submission g…#113451 linesfolded into the change before it
  • fix(planning): show all included atoms in the section readiness card (#1133)#1133139 lines
  • feat: unified section readiness card (Guided Planning center column) (#1132)#11325146 lines
  • feat: planning/review authoring parity — editable planning view in work states …#11312196 lines
  • feat: Curated starter-template layer for the document composer (#1129)#1129736 lines
  • feat(workspace): Planning-stage document & organizer authoring on the shared ou…#11285841 lines
  • fix(planning): remove duplicate ADD SUPPORTING DATA tray (follow-up to #1126) (…#112766 lines
  • docs(solutions): optimistic-reparent unmounts inline-feedback surface (Section …#1122101 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: The Honest Gap Surface — one section lever index (#3 + #5 + #1b) (#1126)#11263273 lines
  • feat(workspace): re-skin shared outline to v3 visual language (#1125)#11251099 lines
  • style(workspace): consistent surfaces — grey/beige left rail, white copilot, ac…#112410 lines
  • refactor(workspace): unified WorkspaceShell — lifecycle-wide panel parity (#112…#11231857 lines
  • feat: The Section Blueprint — A→B→C parts with vault atoms mapped per part (#11…#11212598 lines
  • docs(solutions): reopen-edge novel (state,marker) combination learning + CONCEP…#112076 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: collapse foundation-LOI intake into the planning stage (#1117)#11172193 lines
  • feat(generation): add universal investment-framing principle to generation skil…#1119226 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(workspace): tidy stage dropdown — drop redundant state chip, right-align + …#111626 lines
  • <redacted>#111353 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: Lifecycle phase dropdown (stage navigator) with fluid reopen (#1115)#11152512 lines
  • feat: Honest two-stage planning/review round-trip (#1114)#11142875 lines
  • fix: make foundation-LOI Proceed-to-generation vault-aware (#1112)#1112116 lines
  • fix(workspace): enable document composer in the review planning-view (#1111)#111185 lines
  • feat: funder resolution decoupled from guidelines, intake routed by lookup outc…#11101636 lines
  • feat: flexible vault-backed document builder (U1–U8) (#1109)#11094872 lines
  • docs(plan): mark CI-preflight implemented; reassign DB-free vitest subset to Gr…#11086 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: no-NOFO LOI-first foundation intake (U1–U7) (#1106)#110620463 lines
  • fix(workspace): keep Confirmations controls actionable in the review-planning v…#110562 lines
  • docs(solutions): multi-document surface drops item scope (bare section_id mis-w…#1103100 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: guided-center what's-needed block + Confirmations node (#1104)#11042358 lines
  • chore(workspace): remove rails superseded by the unified document tree (U8) (#1…#11021167 lines
  • docs(solutions): optimistic-mutation popover close-decision pattern (React 19) …#1099184 linesfolded into the change before it
  • feat(workspace): unified grant document navigation — one two-level document tre…#11004037 lines
  • docs(grant-board): mark Phase D plan completed + capture Codex MCP encoding-cor…#1101116 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(grant-board): Phase D U9/U10 — board is the editable grant brief (#1098)#10981166 lines
  • fix(build): NEXT_PHASE-gate the rate-limit-storage production assertion (#1097)#109771 lines
  • feat(grant-board): re-scope hypothesis lever to user-entered primary_hypothesis…#10941386 lines
  • docs(solutions): generalize draftSections writer-registry convention to cover c…#1096147 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(ci): local preflight — verify command + tiered pre-push hook (#1095)#10951659 lines
  • feat(workspace): document-scope copilot "Approve all" batch endpoint (Phase 2b)…#10934999 lines
  • fix(quick-win): reviewEmailSentAt marker + operator-review-email re-pick (#873)…#10928446 lines
  • fix(drizzle): restore missing 0104 snapshot to repair migration lineage (#778) …#10918103 lines
  • fix(documents): observable document/uploaded send drop (#846 pt1) (#1086)#108665 lines
  • fix(quick-win): tighten validateTeaser false-pass/false-fail (#875) (#1090)#1090259 lines
  • refactor(grant-board): levers-only board + compact expand-chips (#1089)#10891453 lines
  • feat(workspace): document-scope copilot + multi-section proposals (Phase 2a) (#…#10882467 lines
  • feat: Grant-level settings board — steering levers + cross-grant-aware editing …#10875128 lines
  • test(drafts): FU-3 commitment-context coverage + selector-form regression guard…#108599 lines
  • chore: safe-tier issue-batch cleanup (phases A–D, 9 issues) (#1084)#10843941 lines
  • feat(workspace): section-scope copilot follow/pin + flag-scoped prompt (Phase 1…#10831243 lines
  • feat: Vault-aware submission packet in the Outline (#1082)#10824159 lines
  • docs(brainstorm): grant control board ("augmentora") chips on planning page (#1…#1067356 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: land stranded brainstorms/plans + concept/solution updates (#1081)#10812115 lines
  • docs: per-section AI-lock + revision-input learnings (from #1062) (#1080)#1080219 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: capture Tailwind-v4 double-translate dialog gotcha + move-field empty-tar…#1075111 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(workspace): loosen resize seam floors so it has real travel (#1074)#10744 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(nofo): demote the NOFO confirmation wall at the creation flow (#1066)#10661792 lines
  • feat: Resizable workspace panel split (writing surface ↔ copilot) (#1068)#10681541 lines
  • feat: Planning↔Review stage switcher + per-section AI lock (#1062)#10624725 lines
  • docs(solutions): capture unified-overlay shared-error-surface pattern (#1065)#106594 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(workspace): legible outline rail + unified add affordance (#1063)#10632816 lines
  • fix(portal): center portal dialogs — drop double translate that pushed them hal…#10616 lines
  • feat: Move a field between finished documents, non-destructively (D2) (#1054)#10545089 lines
  • feat: pre-generate guided surface (Planning v3) (#1053)#10537175 lines
  • feat(layout): add platform-admin dashboard link to sidebar (#1052)#1052123 lines
  • feat: Persistent, editable document names (D-DOC-TITLE) (#1051)#105110182 lines
  • feat: Count and enforce submission limits (D3) (#1050)#10503246 lines
  • docs: land planning/brainstorm/discovery docs stranded on chore/repo-tooling (#…#10493098 lines
  • docs(solutions): add silent-wedge architecture pattern + cross-link fu17 cold-n…#103989 linesfolded into the change before it
  • docs(solutions): structural propose-confirm vault-mutation learning (#1029)#1029109 linesfolded into the change before it
  • docs(solutions): re-runnable materializers merge prior state by stable key (#10…#104448 linesfolded into the change before it
  • docs(plans): packet readiness planning placement direction (#1035)#1035233 linesfolded into the change before it
  • docs(solutions): coverage derivation keys atoms per-requirement, not section-le…#1034103 linesfolded into the change before it
  • docs(solutions): mock @/lib/db/schema by spreading importOriginal (#1030)#103083 linesfolded into the change before it
  • docs(solutions): capture post-edit coherence detector-parity learning (#1027)#102771 linesfolded into the change before it
  • docs(solutions): provenance_unlinked false-flag learning + chain prior instance…#103795 linesfolded into the change before it
  • docs: capture project-lifecycle vocabulary and the modal setState-in-effect pat…#104875 linesfolded into the change before it
  • docs(maps): code-grounded system map of lifecycle phases and levels (#1043)#1043241 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: submission-target document separation — frontend/UX (#1047)#10473954 lines
  • feat(workspace): capture revision instructions in a pre-generation overlay (#10…#10461129 lines
  • refactor: delete orphaned client state-transition guard (#1045)#1045393 lines
  • fix(audits): terminal-state wiring + guarded recovery for stranded extraction_a…#10422043 lines
  • feat: separate documents from submission targets (#1041)#10413868 lines
  • docs: land CONCEPTS.md shared domain vocabulary + surface it in CLAUDE.md (#104…#104092 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(marketing): hero portrait in laptop bottom-left corner (#1036)#1036860 lines
  • fix(ingestion): resolve program scope for logic_model docs + harden silent wedg…#10381493 lines
  • feat(marketing): widen hero laptop + circular portrait (#1031)#103112 lines
  • docs(solutions): provenance stamp-at-origin / carry-through-promote-demote patt…#103391 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(coherence): provenance_unlinked flag for null-source-FK atoms (#1032)#1032998 lines
  • fix(marketing): unblank landing hero images (middleware + real portrait) (#1028)#102813 lines
  • feat: Grantsville landing-page design polish (#1026)#10261309 lines
  • fix(coherence): figure-guard parity on the post-edit coherence pass (#1025)#1025828 lines
  • chore(ops): add resend-document-uploaded recovery script (#956)#95643 lines
  • docs: capture funder knowledge mutation safety pattern (#1024)#1024160 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): funder knowledge capture (Phase 2B) (#1022)#10222365 lines
  • feat(planning): Planning page v2 — per-section requirement coverage (#1021)#10212238 lines
  • chore(vault): tighten provenance_class to NOT NULL (plan U5) (#1020)#10208191 lines
  • fix(scripts): classify backfill atoms from history evidence, add tiebreak gate …#1019590 lines
  • feat(feedback): close the loop with dormant status emails, resolution surfacing…#10179397 lines
  • fix(dashboard): remove duplicate "need you this week" lead line (#1016)#10167 lines
  • feat(vault): provenance class + numeric-conflict readiness advisory (#1015)#101521140 lines
  • fix(workspace): keep packet readiness out of review rail (#1014)#101421 lines
  • chore(repo): add PR template and gitignore basic-memory tooling state (#957)#95720 lines
  • feat: Capture intake user assertions into the vault (Phase 1) (#1008)#10083387 lines
  • feat(workspace): add packet readiness to planning (#1013)#1013854 lines
  • fix(workspace): keep add menus in viewport (#1012)#101249 lines
  • feat(feedback): agent self-flag graduation + triage signal (Phase 3, batch 2) (…#100921718 lines
  • test(drafts): guard grant packet provenance contracts (#1010)#1010296 lines
  • feat(feedback): tenant status/reply tools + operator read-tool wiring (Phase 3,…#1007990 lines
  • test(scripts): harden grant packet shape census (#1004)#1004690 lines
  • docs(solutions): substrate-audit regen convention + propose-confirm agent mutat…#1006106 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: in-app feedback system — Phase 2 (status and triage) (#1005)#100527868 lines
  • feat(workspace): remove user-added outline content (#1003)#1003918 lines
  • feat(workspace): show user document completeness (#1002)#1002135 lines
  • feat(items): exclude user-added readiness blockers (#1001)#1001131 lines
  • test(items): pin U13 user-added Alignment exclusion (#1000)#10008 lines
  • test(drafts): pin U12 funder commitment isolation (#999)#99911 lines
  • test(items): pin null char limit field readiness (#998)#99823 lines
  • feat(deploy): gate Railway deploys on the green main canary (plan 004 U7) (#997)#997119 lines
  • refactor(ci): consolidate to one workflow; full pipeline gates merges (plan 004…#992974 lines
  • fix(ci): ci-skip verifies the PR diff is actually doc-only (#990)#99085 lines
  • feat: in-app feedback system — Phase 1 (the human loop) (#986)#98623422 lines
  • fix(review): harden portal-outline U11-U13 seams from multi-agent review (#989)#989250 lines
  • docs(design): add canonical design.md, slim CLAUDE.md Design pointer (#988)#988450 lines
  • fix(tests): widen narrow draft-sections-schema mocks to importActual+spread (#9…#98745 lines
  • feat(portal-outline): field-type intake, generation isolation, origin-based Ali…#9842048 lines
  • feat(portal-outline): append sections to existing items (#983)#9838325 lines
  • feat: add portal outline document containers (#982)#9822487 lines
  • feat(completability): add remediation actions (#980)#9801987 lines
  • feat(analytics): add nofo_merge_settled_status_dropped event (Half B clock) (#9…#96766 lines
  • feat(scripts): compliance-checklist requirement_identity backfill dry-run harne…#969672 lines
  • feat: add portal outline creation menus (#979)#979374 lines
  • fix: gate stale form flags in cert policy (#977)#977392 lines
  • feat: wire item-scoped portal outline sections (#978)#9788215 lines
  • fix(forms): clear stale form advisory on successful extraction (#976)#976249 lines
  • feat: add portal outline substrate (#973)#9731459 lines
  • feat(forms): preserve stale form materializations (#972)#9727323 lines
  • Materialize extracted AcroForm fields into items (#970)#9701065 lines
  • feat(completability): surface derived packet read model (#966)#9661373 lines
  • feat(packet): add grant packet extraction substrate (#964)#96423221 lines
  • fix(drafts): stabilize compliance checklist identity (#963)#9631116 lines
  • fix(drafts): route derived founding ages to stable source remediation (#962)#9621041 lines
  • fix(workspace): heal WorkspaceLeftColumn test for SubmittedConsole model prop (…#9614 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(workspace): deliverables hub — document→section SUBMITTED console (#959)#9592060 lines
  • feat(workspace): collapse left column (#958)#958181 lines
  • docs(solutions): diagnosing inherited broken base-branch CI failures (#954)#954102 linesfolded into the change before it
  • docs(plan): certified LOI coh-num critical closure (#953)#953214 linesfolded into the change before it
  • docs(brainstorm): deliverables hub redesign (#952)#952163 linesfolded into the change before it
  • docs(plan): certified screen payoff flow (brainstorm + plan) (#951)#951451 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan): collapsible workspace left column (#950)#950382 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(drafts): bind coh-num to selector records (#948)#9481431 lines
  • fix(generation): prefer sourced founding-year phrasing over derived age (#947)#9471 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): make unstructured coh-num mismatches advisory (#945)#9451415 lines
  • fix(workspace): keep portal field modal on-screen (#944)#9442 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(stripe): payments-disabled delivery mode (PAYMENTS_ENABLED flag) (#943)#943663 lines
  • fix(vault): tolerate null prog units in derived metrics (#942)#9421346 lines
  • feat(workspace): add portal submission console (#938)#9381319 lines
  • test(drafts): include ref boundary struct flag (#940)#9403 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): add LOI sentence-quality advisory (#937)#9371935 lines
  • feat: unlock certified proposal downloads (#936)#9361797 lines
  • docs: mark reextract diff plan completed (#935)#9352 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix: surface execute-aware reextract ADD buckets (#934)#934651 lines
  • docs: session artifacts (brainstorms, plans, strategy, tooling) + exit-LOI back…#9334522 lines
  • chore: rescued docs + recovery report from worktree/branch cleanup (#824)#8241962 lines
  • docs(brainstorm): re-extraction propose-candidates reframe + read-only diff too…#918288 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(programs): surface near-match duplicates from extraction (#932)#932378 lines
  • fix(planning): prevent duplicate LOI section arcs (#931)#931259 lines
  • docs(plans): plan 005 post-merge corrections (all-locked R9, U4 reality) (#825)#82557 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: add extraction quality census (#930)#9302206 lines
  • fix(ci): pin Railway CLI to 5.2.0 (5.3.x broke railway up token auth) (#929)#92924 lines
  • fix(drafts): repair portal limit budgets (#919)#919123 lines
  • refactor(drafts): remove dead sectionRequirements plumbing (#925)#925378 lines
  • docs(solutions): read-time derivation lights an inert cache column without a ba…#928130 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(analytics): client components must not import the server-laden analytics ba…#9278 lines
  • feat(analytics): PostHog instrumentation hygiene — metadata contract + parity g…#9231703 lines
  • feat(vault): re-extraction propose-candidates — park, never overwrite (+ operat…#9202974 lines
  • feat(nofo): surface requirement coverage help (#921)#9211348 lines
  • fix(eligibility): numeric comparator — existence-match → value-check for thresh…#9161926 lines
  • feat(vault): re-extraction flywheel — pre-launch mechanism (logic; integration …#9132206 lines
  • docs(solutions): compound the two-part BYPASSRLS preflight convention (#912)#912158 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore: strategic-plan eligibility backfill + compound learning (#869 follow-up)…#877497 lines
  • docs(audits): bank first NOFO modality census baseline (alive -> proceed) (#911)#911104 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: certify-gate event triad + survey trigger (activation drop-off, core) (#9…#910897 lines
  • feat: NOFO modality substrate census (degeneracy gate, Pillar #4) (#909)#9091739 lines
  • docs(solutions): compound the PostHog group-property write contract (#907)#907246 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(nofo): activate verification-modality classifier + atomic atom-key stalene…#903721 lines
  • feat: PostHog org-group property enrichment + internal-traffic filter (#902)#9021997 lines
  • feat(wizard): flag-resolution wizard UI — single-flag guided loop with propose-…#9041796 lines
  • fix(test): unbreak main heavy CI — stale mocks after #895 and #885 (#900)#9008 lines
  • feat: Flag-resolution wizard — foundation (U1 fire-eligibility gate + U2 nav st…#8981174 lines
  • feat(nofo): verification-modality classifier substrate (U1–U3) (#897)#897933 lines
  • docs(solutions): RLS-exempt table cascade GRANT DELETE + companion CI gates (#8…#893107 linesfolded into the change before it
  • fix: global projects-first lock order (eliminate P1 deadlock class; supersedes …#8951126 lines
  • fix(nofo): address code-review findings (#884 autofix) (#889)#889136 lines
  • feat: show funding ask amount in the dashboard Ask column (#882)#882359 lines
  • feat: lifecycle email activation overhaul — Phase 4 (U12 fallback-rate digest r…#887222 lines
  • feat: lifecycle email activation overhaul — Phase 3 (U9, U10a, U10b, U11) (#885)#8851664 lines
  • feat(nofo): verification-modality axis + categorySource provenance + hardened p…#884938 lines
  • fix(security): classify email/quick-win route in SEC-AUTH allowlist (#883)#8832 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(nofo): compound learning (deterministic override + writer-parity) + requir…#879249 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: lifecycle email activation overhaul — Phase 2 (quick-win teaser + human r…#87815423 lines
  • fix(nofo): deterministic requirement-classification override (existence-eligibi…#8691317 lines
  • feat: lifecycle email activation overhaul — Phase 1 (polish + safety scaffoldin…#8711318 lines
  • feat: Per-grant project names with provenance (seed + user tiers) (#870)#87013811 lines
  • docs(nofo-coverage): U2 post-Lever-B re-score record + corpus re-eval trigger (…#868227 lines
  • feat: Flag remediation front door — per-section flag queue (#867)#8671900 lines
  • fix(nofo-coverage): resolve {{ref}} tokens before scoring (Lever B, Track C) (#…#866713 lines
  • feat: Corrections flywheel capture (Slice 1) (#865)#8651135 lines
  • fix(extraction): route over-cap 990 configs to OpenAI gpt-4.1 (issue #832) (#86…#864346 lines
  • feat: Scalar-numeric vault formatter (system-owned currency/percent figures) (#…#8612921 lines
  • fix(test): skill-cwd guard skips co-located *.test.ts files, not just __tests__…#86215 lines
  • test(projects): update extract-nofo dossier-staging pin for resolved funderEin …#85918 lines
  • feat: Resolve funder_ein on the NOFO-content (PDF + paste) creation path (#857)#8571524 lines
  • feat: Computed gap-driven intake (v1 — augment via the human gate) (#854)#8542274 lines
  • docs: close out cite-flag PR-1 (plan completed + roadmap updated, U3 verified) …#8526 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(agent): curate the grant brief generation-prompt projection (Unit i) (#849)#849561 lines
  • ci(deploy): retry railway API calls on transient backboard timeouts (#850)#85036 lines
  • fix(drafts): EDM ref-span gate leaks own-program value before a sentence period…#848112 lines
  • feat(drafts): thread ref-token provenance into citation detectors (cite-flag PR…#8441457 lines
  • feat: Document Ingestion Recovery (manual remove + loud safety net) (#843)#8432226 lines
  • revert(voice): generation-side criteria conditioning (U1+U3+U4 of plan 005) (#8…#8351873 lines
  • docs(roadmap): record prog.* structuring (#838) + unblocked cite-wall closure (…#8395 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): edited-section provenance discriminator convention (#834)#834126 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: prog.* Phase 2 — U4 gate PASS, U5 prefetch render, U6 de-lock script (#83…#8381801 lines
  • fix(extraction): carve prog record_array into a 2nd measuredCall (union-param l…#833839 lines
  • feat(workspace): offered "Revise this section now" fallback (U7) (#830)#83067 lines
  • feat: deterministic revise trigger floor for draft refinement (#828)#8281310 lines
  • docs(voice): conditioning-measurement tooling + runbook (plan 005 follow-up) (#…#819836 lines
  • fix(scripts): U4 model stamp expects OpenRouter's recorded Haiku 4.5 slug (#821)#82124 lines
  • feat: prog.outputs/outcomes live-pipeline structuring (Phase 1 + U4 instrument)…#8183554 lines
  • docs(conventions): a declared shape is not evidence the live write-path produce…#81785 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(voice): condition generation on derived voice criteria (Track 1) — U1+U2 (…#8151542 lines
  • docs(conventions): a new items writer must clear THREE fail-closed registries (…#81612 linesfolded into the change before it
  • docs(brainstorm-seed): bank answered realized-value + lock-semantics fork (#814)#81468 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: Post-generation program reassignment invalidation (#813)#8132122 lines
  • docs(ci): land brainstorm + plan for the CI parallelization restructure (#812)#812383 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: U1 NULL-compound cohort census result — U3 superseded (a=0) (#809)#809275 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • refactor(ci): parallelize heavy pipeline into static ∥ db lanes under a ci aggr…#808527 lines
  • chore(voice): R12 judge-vs-deterministic validation read + first-run results (#…#811183 lines
  • fix(ci): temporary timeout bridge 12->14 to stop at-cap flake (#806)#80616 lines
  • feat: Closed-set program selection in LOI intake (#804)#8041558 lines
  • docs(voice): U6 voice-judge-baseline runbook (judge validation, not recalibrati…#807114 linesfolded into the change before it
  • feat: selector-schema steady-state structuring — U0 gate + U2 read-hardening + …#8032221 lines
  • docs(solutions): runtime asset reads must resolve from process.cwd(), not __dir…#80193 linesfolded into the change before it
  • docs(solutions): items.draftSections writers trip two fail-closed registries (#…#796155 linesfolded into the change before it
  • feat(voice): U5 calibration substrate for the hybrid voice judge (#802)#802824 lines
  • docs(solutions): a more faithful LLM judge can narrow discrimination, not widen…#795103 linesfolded into the change before it
  • fix(inngest): load skill prompts from process.cwd() so the voice judge runs in …#794172 lines
  • fix(drizzle): repair corrupt snapshot chain so db:generate works again (#792)#7926341 lines
  • feat(voice): U4 — render hybrid judge band + per-criterion checklist (plan 007)…#793612 lines
  • feat: cite-attribution grounding for string-array & value_text stat records (D7…#7911100 lines
  • docs(brainstorms): record post-#781 production finding + elevate string-array m…#78916 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(voice): U3 — hybrid fusion + readiness-atomic verdict persistence (plan 00…#7901048 lines
  • feat(voice): U2 — voice judge Inngest fn + event topology (hybrid judge, plan 0…#7861419 lines
  • docs(solutions): capture rename deploy-outage + refresh ci-cd-baseline (#788)#788184 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(voice): U1 follow-ups — re-extraction verification tools + plan/brainstor…#785189 lines
  • docs: rescue unsaved planning + learning artifacts from fix/f1-cleanup (#784)#7842190 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: document upload — text formats (.md/.txt) + instrumented guided rejection…#783947 lines
  • docs: land cert/quality-gate planning lineage on main (provenance for #781) (#7…#7821277 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(voice): U1 deterministic core — criteria-derivation transform (hybrid judg…#7801668 lines
  • feat(drafts): cite-missing-attribution emit policy — suppress own-program groun…#781552 lines
  • fix(brand): sidebar logo monogram BT to G (#779)#7792 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(vault): document + pin the null-vs-absent-key routing divergence (#419, Opt…#77278 lines
  • test(drafts): tighten realSelectorExistsInSchema mock typing (#486) (#770)#7707 lines
  • feat: rename Boon Town to Grantsville (grantsville.co) (#775)#775683 lines
  • feat: extraction coverage-gap telemetry (#777)#77713690 lines
  • chore(scripts): expose --extraction-success-rate-min/-max flags on dispatch-mig…#769177 lines
  • fix(auth): disable org member-invitation creation at the source (V1 non-goal) (…#776212 lines
  • fix(ci): drop strict mode, add main canary to kill the merge-rerun cascade (#77…#774481 lines
  • fix(agent): strip array .max() from submitRevision tool schema (#290) (#763)#76313 lines
  • docs(conventions): canonical home for the two-role DB model (#667) (#771)#771120 linesfolded into the change before it
  • docs(brainstorms): issue-backlog grouping + triage requirements (#773)#773116 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(cert): Voice Authenticity advisory at Tier 1 (scored-and-shown, non-blocki…#768944 lines
  • docs(drafts): preserve flag-leak brainstorm, plan, and production probe (#760)#760435 lines
  • fix(drafts): strip pipeline-owned atom_ref sub-fields from LLM contract (#759 n…#761717 lines
  • fix: Upgrade better-auth to 1.6.11 to close the magic-link session-minting race…#7621235 lines
  • fix(drafts): omit flags from generation schema to close flag-leak vector (#759)#759387 lines
  • docs(solutions): trustworthy fixed-vs-pre-fix audit re-run via snapshot-only (#…#758232 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): F1 cleanup — resolve 4 critical hand-named flags + brainstorm doc …#756681 lines
  • feat: Advisory-vs-blocking framework v1 — encode per-family cert-gate policy (#…#7572075 lines
  • fix: re-bind mismatched record_array selector indices in generation (#755)#7552662 lines
  • docs(audits): L4 calibration plan status active->completed (U-final, verified l…#7513 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: prevent structurally-malformed Class-B atom_refs at the write boundary (#…#7541539 lines
  • fix: Backfill program_id discriminator on historical prog.* atom_refs (F-atomre…#7532189 lines
  • <redacted>#752847 lines
  • fix(audits): fold coordinator-row minting into per-layer extraction-audit CLIs …#750660 lines
  • fix(audits): repair compound-atom placeholder-menu resolved_value (F-atomrefs A…#7491817 lines
  • docs(audits): Layer 4 (atomRefs trace) first calibration — Gate 4 + verdict pas…#748669 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(audits): L3 classifier — remove discarded-justification cap (F-classifier) …#746274 lines
  • fix(audits): L3 Track B — neutralize hyphen-range spurious negatives (F-hyphen-…#747347 lines
  • fix(audits): L3 alt-impl — stop provable-arithmetic over-fire (F-altimpl U1) (#…#745461 lines
  • docs(audits): resolve capstone 18/19 mirror_diverges observation — two scoped f…#7444 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): F-track-b status active->completed (verified live, #742 merged) (#…#7432 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(audits): L3 Track B — strip ref tokens before numeric extraction (F-track-b…#742522 lines
  • feat(audits): compliance-surface finding-key harmonization to honest `requireme…#741498 lines
  • docs(audits): Layer 3 (derivation-classifier) Gate 3 calibration — first real d…#740718 linesfolded into the change before it
  • docs(audits): L1 Gate 1' cycle-close operator log (terminal re-run capstone) (#…#7391 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): F-mirror-id-resolution status->completed + operator-log cycle note…#7384 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(audits): F-mirror-id-resolution — L1 coverage mirror keys by requirement t…#7371010 lines
  • docs(audits): F-json U5 — Layer 2 closure operator log + plan completion (#735)#735153 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(audits): F-overextract — skip workflow requirements from L1 coverage class…#736583 lines
  • fix(audits): namespace-qualify L1 coverage live-atom key set (F-vault-lacks-nam…#734544 lines
  • fix(scripts): backfill-990-json-chunks takes --org-id (was silent RLS no-op on …#733302 lines
  • docs(plans): ship L1 (Coverage) calibration plan at status completed (#732)#732444 linesfolded into the change before it
  • docs(solutions): recursive corollary on substrate-claim discipline (reviewer se…#72985 linesfolded into the change before it
  • docs(audits): Layer 1 (Coverage) first-real-data Gate 1' calibration on Connect…#731238 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(audits): F-json — Form 990 ProPublica JSON text-chunking for Layer 2 subst…#7301421 lines
  • docs(solutions): LLM prompt iteration discipline (hold-out fixtures + verdict r…#725231 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(lifecycle-emails): minimal body-link unsubscribe in editorial layout (#728)#72864 lines
  • docs(solutions): compound two learnings from PR #699 writer-side atomRefs activ…#727262 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): mark atomref writer-side activation plan as completed (#726)#7263 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): held-out fixtures + cross-bucket regression patterns for LLM i…#724199 linesfolded into the change before it
  • feat: Layer 2 deterministic-check date format normalization (F-date) (#723)#723999 lines
  • ci(inngest): plan-tier concurrency cap gate (#721)#721297 lines
  • docs(solutions): Inngest plan-tier concurrency silent sync-batch failure (#722)#722142 linesfolded into the change before it
  • docs(solutions): bundle 4 solution docs from recent sessions (2026-05-23) (#719)#719546 linesfolded into the change before it
  • feat(u6f): activate NOFO_CONDITIONAL_TAXONOMY_MISS event (#716)#7162196 lines
  • copy(lifecycle): six register-tightenings across welcome + anchor + stalls (#72…#72029 lines
  • chore(plans): flip 2026-05-23-001 F-num/F-shortnum plan status to completed (#7…#7183 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: Layer 2 deterministic-check numeric format normalization (F-num + F-short…#7171070 lines
  • feat: Layer 2 provenance-granularity decision — ratify deterministic_containmen…#715617 lines
  • feat(vault): U6g re-promote funder.* atoms to CRITICAL (post-U6i verified) (#71…#71459 lines
  • fix(u6i): skip archived-only GHGF projects at enumeration boundary (#713)#71330 lines
  • feat: U6 conditional eligibility PR 2 (evaluator + telemetry + migration) (#712)#7122653 lines
  • feat: U6 conditional eligibility PR 1 (schema + U5 reversal — partial) (#711)#7111756 lines
  • feat(vault,u5): conditional eligibility reader sweep + readiness funder-scope (…#7042399 lines
  • docs(conventions): two conventions distilled from the U2+U3+U4 bundle review (#…#705319 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): characterization-first gate catches wrong defect framing (#702)#70272 linesfolded into the change before it
  • P0: deletion email recovery URL must carry recovery_view token (not cancel_dele…#703258 lines
  • feat(agent): list_atom_ref_strict_rejections tool + generation skill update (#7…#701238 lines
  • feat(vault): U2+U3+U4 funder.* namespace + funder_ein scope (bundle) (#700)#70013139 lines
  • fix(nofo): two-domain merge on re-confirm rebuild (plan U1) (#696)#6961480 lines
  • feat(audits): Layer 3 soft-fail v2 + first-real-data findings + compounded lear…#695545 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: writer-side atomRefs substrate fix — Activation PR (U4-U14) (#699)#69920487 lines
  • docs(audit): bank strategy-code alignment audit chain (2026-05-20) (#694)#6941050 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): document Zod 4 .uuid() version + variant bit strictness (#693)#693161 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: writer-side atomRefs substrate fix — Capability PR (U1-U3) (#692)#6921718 lines
  • docs: cert-gate brainstorm + spike falsification + conditional-flattening follo…#691388 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: Voice Authenticity canonical-axis governance + generation.md weight refer…#690451 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(strategy): defer Federal Proposal tier to post-V1 (#689)#68919 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(audits): Layer 3 soft-fail on classifier nonconforming output (#687)#687603 lines
  • docs(brainstorm,plan): voice fingerprint schema v1.2 spec edit (#688)#688254 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(audits): preserve JSON null literal on jsonb NOT NULL snapshot writes (#685)#685317 lines
  • feat(audits): wire Layer 3 production classifier (#684)#684453 lines
  • feat: extraction-quality audit subsystem (U1–U9 + /ce-code-review hardening) (#…#68326124 lines
  • docs(solutions): compound two PR-2 learnings — activation substrate vs UI surfa…#682280 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • refactor(pr-2): trim onboarding UI surface — keep API route as activation subst…#681485 lines
  • feat: board+staff demographics taxonomy widening PR-2 (U7 + U8 activation) (#67…#6772547 lines
  • fix(migration): drop dei_statement trim + correct readVaultAtom arg order (#678)#678362 lines
  • docs(solutions): capture two learnings from PR #675 review (#676)#676252 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: board+staff demographics taxonomy widening (PR-1 substrate-less ship) (#6…#6753328 lines
  • feat(deletion): plan-aligned Inngest retention via INNGEST_EVENT_RETENTION_DAYS…#674130 lines
  • Phase 5 close-out: fix audit-cascade-rls-policies.ts + pre-committed rollback t…#673112 lines
  • docs(claude.md): require draft-by-default for PRs (#672)#6721 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(brainstorm): resolve gov.diversity_table row-shape question (#670)#670444 linesfolded into the change before it
  • ci: PR-time smoke + draft-skip full CI to cut Actions minutes (#671)#671209 lines
  • feat(drafts): needle-in-atom coverage-revision prose conditional (#669)#6691313 lines
  • Phase 5 close-out (3/3): drill exit-code fix + activation runbook sync (#663)#663187 lines
  • docs(solutions): refresh 2 migration-ordering docs for the preDeployCommand rea…#66617 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Phase 5 close-out (2/2): deletion API flag-gating + nav + missed-send alert dea…#662344 lines
  • Phase 5 close-out (1/2): FU-18 verification + cascade rollback SQL (FU-26) (#66…#661927 lines
  • fix(resolve-eligibility): accurate log severity for the prog-key strip (#665)#665297 lines
  • docs(solutions): compound the Railway pre-deploy migration role failure (#664)#664101 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix: re-score eligibility on vault atom write (#659)#6591943 lines
  • docs(deploy): correct preDeployCommand comment — real cause was DB role, not np…#66018 lines
  • docs(solutions): document the forked Drizzle migration snapshot-chain failure (…#657140 linesfolded into the change before it
  • fix(deploy): invoke drizzle-kit directly in preDeployCommand, not via npm (#658)#6589 lines
  • feat(revise-section): thread evidence_summary into coverage revisions (#656)#6561248 lines
  • feat: completion-email missed-send reconciliation sweep (FU-16(c)) (#650)#6507830 lines
  • fix: relocate stale op:// references in env template (#652)#65227 lines
  • fix(coherence): selector-carry-over guard (certify-one-LOI cycle, U1-U7) (#651)#6511929 lines
  • docs(solutions): how to verify the token-consistency retry from telemetry (#648)#648114 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(deploy): run db:migrate on Railway deploy via preDeployCommand (#649)#649105 lines
  • fix(drizzle): repair forked migration snapshot chain (0068-0072) (#647)#6474529 lines
  • feat: in-generation token-consistency retry loop for orphan {{ref}} tokens (#64…#646699 lines
  • docs: plan the in-generation token-consistency retry loop (Option A) (#645)#645312 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: record token-inconsistency followups and reinstate plan 002 (#644)#644414 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(followups): consolidate Phase 4 deletion follow-ups (FU-1..29) (#643)#643127 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Revert "fix: route loi_project section generation to openai/gpt-5.4 (#641)" (#6…#642245 lines
  • FU-22 rollback-drill fixtures + PR-D Finding A — cascade allowlist net alignmen…#6377158 lines
  • fix: fail fast when DATABASE_URL is unset (#640)#64016 lines
  • docs: commit STRATEGY.md to main (#639)#63980 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix: route loi_project section generation to openai/gpt-5.4 (#641)#641245 lines
  • fix: voice extraction hardening — PR #202 review follow-ups (#205, #206, #207, …#638998 lines
  • fix: EDM follow-ups from PR #429/#442 review (#432, #433, #434, #435, #446) (#6…#6361579 lines
  • fix: stop the start-new-grant surface jumping when input-mode tabs switch (#635)#63544 lines
  • docs(solutions): compound — name the canonical snapshot a verification claim is…#634133 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(agent): correct stale TokenInconsistencyError reference in generation.md (#…#6332 linesfolded into the change before it
  • feat: U10 — telemetry instrumentation (plan-007 final unit) (#630)#6301367 lines
  • docs(solutions): compound the readiness-gate empty-text bug (#626) (#629)#629113 linesfolded into the change before it
  • docs(plan-007): U10 telemetry verification report + deepened spec (#628)#628872 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix: orphan-ref degrade-and-recover observability follow-ups (#632)#632315 lines
  • fix(coherence): degrade Layer 3 gracefully on per-call LLM timeout (#631)#631417 lines
  • fix: readiness gate holds empty-text sections not_ready (#626) (#627)#627192 lines
  • fix: make orphan ref tokens non-fatal to LOI generation (#619)#6191198 lines
  • docs(solutions): compound the spec-internal-consistency verification pattern (#…#625162 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): external-API fixture shape must match mapper output (#614)#614114 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: U9 — renewal-collision detection on dashboard-add entry (#620)#6201207 lines
  • feat: U8 — save-to-watch action + target_funders writer/route (#616)#6161810 linesfolded into the change before it
  • docs(plan-007): U9 renewal-collision verification + deepened spec (#618)#618510 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): capture projection-layer tracing learning from #613 (#615)#615117 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(deletion): Phase 4 PR-C — cascade activation readiness (#612)#6129821 lines
  • fix(coherence): preserve array-valued atom values through the projection layer …#613792 lines
  • feat: U7 — dossier zone, always-confirm gate, portalPendingAt lifecycle (#611)#611804 lines
  • chore: remove stray temp file committed in U6.5 (#610)#6101 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix: remove copyable real-key compound-atom example from generation prompt (#60…#60950 lines
  • docs(plan-007): lock FU-5 — lookup-channel sparse-requirements decision (#608)#608316 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: U6.5 — lookup branch → channel stager (plan 007) (#607)#607726 lines
  • docs(plan-007): U7 halt — A2 re-lock + insert U6.5 (confirmation-surface reconc…#606154 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: U6 session learnings — build-gate miss + FU-4 paste-abandonment orphan (#…#605450 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: U6 — funder-name input + paste channel + URL affordance (plan 007) (#604)#6042548 lines
  • feat(deletion): Phase 4 PR-B — +30d hard-delete cascade (U15-U17) (#603)#60311433 lines
  • feat: EIN-pinned funder-lookup orchestrator entry (U5.5c) (#602)#602338 lines
  • docs(plan-007): insert U5.5c — EIN-pinned orchestrator entry (U6 halt #2) (#601)#60156 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): two workflow learnings from the plan-007 U5.5b session (#600)#600186 linesfolded into the change before it
  • docs(plan-007): deepen U6 against U5.5b's shipped substrate contracts (#598)#598102 linesfolded into the change before it
  • docs(plan-007): bank U5.5b deferred code-review findings as follow-ups (#597)#59756 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(projects): U5.5b — project-creation-from-lookup bridge (#596)#5965655 lines
  • docs(solutions): compound — capability/activation split for probe-only opt-ins …#594395 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan-007): U5.5a Q2 writer-universe completeness re-lock — 4-category taxo…#595183 lines
  • fix: remove dead font preloads + hydration mismatch in GrantSummaryCard (#592)#59240 lines
  • docs(plan-007): Q4 substrate-claim correction — re-lock with staging-promote sh…#593147 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(deletion): Phase 4 PR-A — bulkDeleteR2Objects helper (U14) (#589)#589957 lines
  • docs(solutions): two patterns from plan 007 PR #581 session (#587)#587653 linesfolded into the change before it
  • docs(solutions): streaming-response in-band error envelope pattern (PR #585) (#…#588291 linesfolded into the change before it
  • docs(solutions): execution-tier mirror — ce-work surface survey as falsificatio…#59035 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan-007): U5.5 substrate decisions + implementation plan for project-crea…#591536 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(runbook): R11 interpretation guide — diagnostic-not-routing discipline (#5…#586290 lines
  • feat(deletion): Phase 3 — recovery page + cancel + export endpoints (U11+U12+U1…#5857743 lines
  • feat(funder-lookup): start-new-grant flow consolidation + EIN-anchored funder l…#58121556 lines
  • feat(cert-distribution): U5 R11 — per-component voice score CSV emission (#584)#584267 lines
  • docs: vault atom axis purity convention + Pitfall #9 cross-link (#578)#578217 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(voice-cert-gap): U6/U7/U8 — B′-degraded probe substrate + review fixes (#5…#5824322 lines
  • chore(test): remove orphaned hasDb constant in schedule.test.ts (#583)#5832 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(deletion): Phase 2 PR-C — lifecycle email family + reminder cron (U10) (#5…#5807635 lines
  • docs(brainstorms): bank three open seams from payment-rails Step 0 (#574)#57472 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): substrate-finding repair-option routing for one-shot scripts (…#579154 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(deletion): Phase 2 PR-B — schedule path + danger-zone surfaces (U8+U9) (#5…#57512419 lines
  • feat(voice-cert-gap): U1-U5/U9/U11 + 3-batch code-review fixes (#577)#5778478 linesfolded into the change before it
  • feat(planner): route board/staff/diversity requirements to gov.* atoms (#576)#576123 lines
  • docs: NOFO Coverage diagnostic — R22 side-by-side + R23/R24 caveats (#573)#57342 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): NOFO Coverage diagnostic operator-curation disciplines (#571)#571160 linesfolded into the change before it
  • feat(deletion): Phase 2 PR-A — withAuthRejectDeletedEntity + withRecoveryToken …#5721676 lines
  • docs(solutions): coh-num cross-section retirement learnings (compound + refresh…#563164 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: account/org deletion — Phase 1 substrate (U0–U5 + substrate-audit) (#568)#56826253 lines
  • <redacted>#5664238 lines
  • feat(nofo-diagnostic): NOFO Coverage gap attribution procedure (#570)#5703440 lines
  • feat: fiscal-sponsorship onboarding signal (Payment Rails Step 0) (#569)#5691312 lines
  • fix(inngest): drop lifecycle-anchor concurrency 10 -> 5 to match plan tier (#56…#5672 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(onboarding): two unified-lookup UX bugs (layout-shift + Not us behavior) (#…#56566 lines
  • fix(propublica): tolerate missing income_amount + total_employee_count in searc…#56456 lines
  • docs(solutions): compound learnings from PR #553 + PR #557 + review (#562)#562579 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): backfill PR #553 + PR #557 decision artifacts onto main (#561)#561968 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(runbook): fix activation runbook URLs and DNS verification steps (#551)#55121 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: Unified org-lookup input (EIN or name, intelligent fallback) (#557)#5572263 lines
  • fix(coherence): retire cross-section coh-num emit (Check 2) (#555)#555993 lines
  • chore(deps): upgrade posthog SDKs and bump node engines (#556)#55667 lines
  • fix: show non-zero size for small NOFO uploads (B/KB/MB) (#554)#554146 lines
  • fix: Dummy-proof EIN entry (accept 8-digit, friendly errors) (#553)#553430 lines
  • docs(followups): FU-8 — phantom cross-section coh-num flags on narrative array …#552187 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(privacy): add v0 privacy policy + middleware fix for email landings (#550)#550360 lines
  • feat: Phase D+E onboarding lifecycle emails — webhook + digest + flag flip (#54…#54414220 lines
  • docs(solutions): R10 close-out correction — preflight discipline + investigatio…#54899 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): findings-ladder discipline — subsumption vs probe-blocking (#5…#549160 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): refresh operational-changes — add cross-ref to new branch-off-…#5475 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): pre-register falsification-gate artifacts on branch cut off ma…#546120 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(claude-md): add Decision Discipline section with shape-of-urgency check (#…#5456 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(plans): flip two completed plans active → completed (#543)#5434 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(runbooks): U4 outcome — restore F1, reframe F7, structural feedback (#542)#542188 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • <redacted>#54122 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • <redacted>#540422 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(lifecycle-emails): Phase C — user-channel emails (welcome + anchor + stall…#5362071 lines
  • fix(auth): surface Resend magic-link and password-reset failures to client (#53…#53942 lines
  • fix(auth): surface Resend verification-email failures to client (#538)#53811 lines
  • fix(auth): construct Upstash client explicitly in rate-limit factory (#537)#537157 lines
  • docs(phase-b): record PR #534 follow-up routing decisions (#535)#53528 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(lifecycle-emails): Phase B — INTAKE_STARTED + VAULT_POPULATED_V1 emit site…#5345305 linesfolded into the change before it
  • fix(auth): include apex boon.town in Better Auth trustedOrigins (#533)#5331 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: Phase A — Onboarding lifecycle-email substrate (U1, U2, U3, U4, U21) (#53…#53129966 lines
  • <redacted>#532545 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(brainstorm): External Grounding Subsystem (verification + research) (#530)#530216 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(cut3): closeout — R16 verdict GREEN; U4 R9 escalation to Cut 4 (#529)#529179 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(u5): Tier 1 replay outcome — substrate convergence verified, 3 deferred fi…#527321 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(audits): fix Cut 3 U5 surface label (S3 → S4) (#528)#5286 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(audits): Cut 3 U5 Mode-D telemetry observation — R10 fires green (#526)#526100 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): convention tests prove what they enumerate, not what they don'…#52573 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(scripts): Cut 3 U3 scaffolding — Eval Plan dispatch + assertion verifier (…#5152074 lines
  • fix(nofo-shape): widen compliance_requirements.category enum at 5 missed consum…#524152 lines
  • feat(scripts): Cut 3 U3 PR 2 — pre-registered Eval Plan assertions (#522)#52263 lines
  • <redacted>#523442 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • test: remove ci-skip verification marker (verification noted in PR thread)409a1807 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • test: doc-only commit to verify ci-skip path filter (PR #521)4aae4377 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • ci: skip full pipeline for doc-only changesedc5ef051 lines
  • docs(r6-convergence): canonical baseline + 2 anomaly observations + two-falsifi…#520287 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): U2.7 activate producer-side record_array support in getAtomStructu…#517656 lines
  • feat(scripts): PR-E — backfill req_014 not_applicable attestation on 12 Connect…#519461 lines
  • docs: correct FU-X1c PR-A interaction framing per ce-code-review finding #3 (#5…#51313 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): line-416 eligibility-overlay stale-snapshot race (deferred) (#…#514114 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): FU-1/PR-A/B/C/D workstream lessons-learned (#518)#51836 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • <redacted>#516572 lines
  • <redacted>#512702 lines
  • fix(coverage-eval): narrative target_atom_keys clobber (FU-X1b PR-A) (#509)#509332 lines
  • <redacted>#5071938 lines
  • docs: Phase 0 readiness reframing - FU-1 investigation findings (#508)#508562 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(coverage-eval): preserve workflow-typed rows in nofo_requirements write-bac…#51083 lines
  • feat(vault): Cut 3 U1 measurement-plan atoms + U1.5 directLockEligible broadeni…#506693 lines
  • docs(solutions): Drizzle silent-success failure modes — convention (#505)#505256 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(inngest): backfills' load-candidates EXISTS-as-projection bug (#503)#503580 lines
  • <redacted>#504702 lines
  • fix(eligibility): protect not_applicable from silent revert (U1b + U2b follow-u…#502570 lines
  • investigate(inngest): diagnostic v2 — Drizzle-typed select side-by-side (#501)#501175 lines
  • docs(plans): Stage B Cut 3 — Tier 2 narrative generation (Eval Plan forcing fun…#500665 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • investigate(inngest): one-shot RLS query-shape diagnostic (#498)#498219 lines
  • docs(plans): LOI eligibility — plan + pre-implementation P0 pins (#499)#499752 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): LOI eligibility substrate coverage — follow-ups (#496)#496215 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(agent): hoist resolveEligibilityAtomKeys out of confirm_nofo_extraction tx …#495868 lines
  • fix(vault): resolve Next.js 16 dynamic-segment conflict on /atoms route (#494)#494556 lines
  • feat: LOI eligibility substrate coverage — unblock first-LOI-certifies on Conne…#4933108 lines
  • fix(inngest-gate): drop admin/ exemption from Check 2 (declared-without-consume…#49248 lines
  • docs(solutions): split-allowlists-asymmetric-writer-site-inclusion learning (#4…#491179 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): per-atom recently-changed indicator with revert (#490)#49012214 lines
  • docs(solutions): carry-by-ref outcome + two-phase emit design pattern (#489)#489179 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): substrate-inspection cluster meta-doctrine (#484)#484198 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan): close out 2026-05-05-002 tranche (T1 verified) (#481)#48127 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(drafts): activate Mode-D auto-repair (flip enable flag) (#488)#48827 lines
  • docs(solutions): plan-lifecycle pre-dispatch trilogy — new learning + sibling c…#487152 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(drafts): Mode-D auto-repair — synthesize record_array selectors from resol…#4831650 lines
  • <redacted>#48218 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): Mode-D record_array selector inference plan (#480)#480336 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): #283 build-time guard — COMPOUND_ATOM_KEYS ↔ selector_schema avail…#4792202 lines
  • chore(scripts): U4 SAM ed_email/ed_phone adversarial dry-run (#478)#478288 lines
  • chore(plans): follow-up sweep — 4 false negatives flipped (#477)#477431 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(plans): plan-status reconciliation 2026-05-08 sweep (#476)#476312 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plan): close out -001 (shipped in PR #422) (#474)#47476 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(plans): add audit script for stale status: active plans (#475)#475706 lines
  • docs(solutions): LLM-extraction failure triage ladder (workflow-issue) (#473)#473242 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • ops(scripts): manual extraction for 1940f68e (closes #467, R7=10/10) (#471)#471294 lines
  • fix(vault): swap classifier model to openai/gpt-5.4 + revert PR #469 prompt str…#470101 lines
  • fix(vault): C1 residual — required-field reinforcement + per-item rule applicat…#46993 lines
  • fix(drafts): #408 buildVaultGapFlags selector-aware span lookup (#465)#465308 lines
  • feat(vault): C1 activation — register prog.outputs + prog.outcomes (U6) (#466)#46626 lines
  • fix(scripts): filter loadSourceAtoms to unmigrated atoms only (#464)#46493 lines
  • fix(vault): #436 follow-up — exported isRenderSelectorError predicate, AST swee…#463503 lines
  • feat(vault): optional marker for record_array string sub-fields (closes #455) (…#4621722 lines
  • fix(vault): #436 SelectorMissingError consumer cascade audit — sentinel-prefix …#46179 lines
  • chore(plans): mark 2026-05-06-002 as completed (shipped via PR #452) (#460)#4603 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): queue-layer-idempotency learning + 5 substrate-audit-gap refre…#441385 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): strip array .max() from residual classifier schema (Azure ceiling)…#28936 lines
  • docs(solutions): bundle review findings by site, not just by ID, when fixes sha…#45999 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): narrow CompoundValue to discriminated union (PR A) (#458)#4581512 lines
  • feat(vault): C1 LLM-extraction prompt iteration — anti-over-split + non-empty s…#454451 lines
  • feat(scripts): add --pass-a-min and --pass-b-confirm-max flags to dispatch scri…#453176 lines
  • fix(vault): #449 LLM-extraction record_array storage-shape contract (#452)#4522310 lines
  • docs(solutions): CLAUDE.md routing-layer pattern (#450)#450228 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): TypeScript typing-as-regression-guard claims must trace the ac…#451191 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • refactor(claude-md): routing-layer refactor + integrate 8 main-side rules (#448)#4481929 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): C1 U4 — operator-triggered Inngest dispatch + cascade-audit gap cl…#4421187 lines
  • docs(c1): U2 prose walk + schema lock proposal — Decision 1 finalized, four ref…#440260 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(inngest): lookup-990 ingests ProPublica JSON for non-XML orgs (#438)#4381621 lines
  • docs(solutions): static call-site lock as a CI gate pattern + 3 refresh edits (…#387268 linesfolded into the change before it
  • docs(solutions): capture exhaustiveness-witness pattern + cross-surface blind s…#389175 linesfolded into the change before it
  • docs(solutions): routing tag must match verdict substance in ce-code-review syn…#4050 linesfolded into the change before it
  • docs(carries): sharpen multi-org-substrate-verification + correct outcomes-indi…#42713 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore(plan): flip status active->completed for projects.draftSections cleanup (…#4372 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): helper implicit-hedging audit on named-purpose evaporation (#4…#439142 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): C1 U3 — EDM record_array iteration + cascade-audit-surfaced siblin…#429551 lines
  • fix(drafts): cite-* detector false positives in production (#430)#43050 lines
  • docs(solutions): post-PR-424 follow-up — round-2 adversarial learning + parent …#428142 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs: cross-workstream artifacts — U6 substrate-track learnings, PCP U9 selecto…#3975261 lines
  • feat(inngest): close 8 event-system drift instances + ship bidirectional parity…#4243287 lines
  • docs(solutions): four workflow learnings from PR #422 (substrate verification +…#426360 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(observability): emit analytics event on coherence-flag clearance (#421)#421765 lines
  • docs(solutions): two compound learnings from PR #422 (#425)#425218 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat: add contact and year-founded atoms (taxonomy + extraction) (#422)#422357 lines
  • docs(solutions): Radix Dialog translate-centering clips tall content (PR #418 r…#423113 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): SelectorMissingError — read-time tolerance for null-valued typed s…#416386 lines
  • fix(cut-1): modal content overflowing past viewport top (#418)#4182 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): operational recipe for applying drizzle migrations OOB via Neo…#420172 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): drizzle migration-number collision rebase-recovery recipe + ba…#417173 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(items): Stage B Cut 1 — + flow activation (PR B) (#415)#4154263 lines
  • feat(vault): lift writer-side admission gate for record_array in migrate-mode L…#414284 lines
  • feat(projection): U6 — retire coh-def-* legacy flag class (#403)#4031252 lines
  • feat(vault): per-document upload reconciliation report (#409)#4096156 lines
  • fix(drafts): skip record_array fields in EDM compound-atom token-set synthesis …#412159 lines
  • docs(solutions): EDM × locked compound atoms — synthesis bug + data-class drift…#411258 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(db)!: items composite-FK precondition (Stage B Cut 1, U1) (#410)#4105655 lines
  • feat(drafts): Mode-A auto-repair — unblock LOI org_credibility + budget section…#407575 lines
  • docs(brainstorms): architectural depth vs launch readiness — pre-launch posture…#406227 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(drafts): annotate compound atoms with legal selectors in prompt inventory (…#404207 lines
  • fix(drafts): unblock LOI generation — Anthropic rejects integer min/max (#402)#40216 lines
  • feat(db)!: drop projects.draft_sections column + complete synthesis-function se…#3985516 lines
  • feat(projection): U5 — coh-paraphrase-entailment + narrative-deferred widening …#3923925 lines
  • feat(drafts): structural-tell + citation-presence detectors (U1-U4) (#401)#4012926 lines
  • feat(v1-manifest): U8 — lint guards + active-row soft-archive convention (#396)#3961313 lines
  • docs(solutions): add Drizzle commit-on-resolve / partial-commit owned-tx patter…#391390 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): zod-literal fixture must import constant convention (#395)#395168 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(v1-manifest): U7 — Coverage and Alignment scope filters (R10, R11) (#393)#393233 lines
  • fix(analytics): enable $pageleave + $web_vitals capture and reverse proxy inges…#39423 lines
  • feat(v1-manifest): U6 — items.readiness derivation + R9 SUBMITTED/CERTIFY gate …#3901719 lines
  • feat(projection): U4 — coh-paraphrase-hallucination + coh-narrative-deferred em…#3882494 lines
  • feat(v1-manifest): U5 — SQL outliers retarget + lock-predicate sub-tasks + defe…#3841055 lines
  • test(projection): U3 handler harness — close #379 (a)(i)-(vi) (#382)#382786 lines
  • feat(projection): U3 — pinned-token authoring at atom-save time (#381)#3811229 lines
  • feat(v1-manifest): U4 — wide writer/reader retarget projects.draftSections → it…#3781867 lines
  • docs(solutions): record searchOffsets-invariant U2 recurrence in vault-scope-as…#37720 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(v3a): add CLAUDE.md Never Do entries for sanitizer + actor invariants (#31…#3152 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(projection): U2 — coh-snapshot-stale-* emitter for atom-to-section project…#3761380 lines
  • chore(devx): track .env.local.op-template + launch.json for worktree-aware dev …#372119 lines
  • feat(v1-manifest): U3 — items-aware accessors + R2a invariant validator (#373)#373557 lines
  • docs(solutions): drizzle migration tracking timestamp-ordering footgun + devx (…#374236 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): AgentFeedPanel auto-scroll smooth-scroll race learning (#367)#367560 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(projection): U1 — atom-to-section projection contract substrate (#370)#3707180 lines
  • feat(v1-manifest): U2 corrections — archive carry-through + readiness floor + p…#3695110 lines
  • feat(v1-manifest): U2 — items backfill migration + halt-on-mismatch verificatio…#3684564 lines
  • feat(v1-manifest): U1 — items table substrate + R7a invitation column (#363)#36313794 lines
  • fix(workspace): AgentFeedPanel auto-scroll race during streaming agent reply (#…#36646 lines
  • feat(v3b): PR-E2 — Pattern A sub-chip wiring + activation flag flip (#365)#3651289 lines
  • feat(v3b): PR-E1 — Pattern A parent surface substrate (no flag flip) (#364)#364694 lines
  • docs(solutions): railway up --ci does not auto-inject RAILWAY_GIT_COMMIT_SHA (#…#344137 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(v1-manifest): commit manifest items substrate plan + supporting learnings …#3621417 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): Drizzle SAVEPOINT pattern for composite handler atomicity (#36…#361227 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): V3b workflow discipline learnings — starter-prompt substrate v…#360230 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(v3b): PR-D3-followup — composite create_atom_and_cite body + U21 wiring + …#3592103 lines
  • docs(solutions): two-flag activation gate pattern for incomplete sub-features (…#358180 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(v3b): PR-D3 — create_atom_and_cite substrate landing (stub + classifier ro…#357790 lines
  • docs(v3b): commit pattern-a-activation plan (#356)#3561617 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(v3b): PR-D3a — applySpliceInTx extraction + add_citation_token_splice hand…#3552795 lines
  • feat(v3b): PR-D2 — CiteCreateModal + chip-dispatch sub-chip request shape (#354)#3541168 lines
  • feat(v3b): PR-D1 — Add citation cite-existing path (substrate) (#353)#3531632 lines
  • feat(v3b): PR-C — Remove number splice + telemetry window tracker (#350)#3501192 lines
  • docs(solutions): capture ESLint warning ratchet convention from 188→0 cleanup a…#352195 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): capture React Compiler lint cleanup gotchas + cross-link to la…#327145 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): substrate audit vs code review correctness coverage boundary (…#351179 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(lint): resolve 9 react-hooks/exhaustive-deps + flip ratchet to 0 (closeout)…#349457 lines
  • refactor(lint): sweep 25 test files + 1 stray prod warning, ratchet 51→9 (plan-…#34890 lines
  • feat(v3b): PR-B — Mark unprovenanced (schema + emitter suppression + handler) (…#34610553 lines
  • refactor(lint): sweep 64 prod files clean of no-unused-vars + ratchet 154→51 (p…#347247 lines
  • fix(lint): honor ^_ in args/catches + ratchet to 154 (plan-003) (#345)#345484 lines
  • feat(v3b): PR-A — V3a-medium-deferred substrate extension for Pattern A activat…#3422013 lines
  • docs(solutions): session-bound branded IDs as structural defense for capability…#314217 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(ci): pass commit message to `railway up` via -m (#343)#3437 lines
  • fix(ci): inject RAILWAY_GIT_COMMIT_SHA before `railway up` (#341)#34124 lines
  • feat(ci): Railway deploy gate + /api/health commitSha + push:main failure issue…#340153 lines
  • docs(ci): complete runbook stubs + CLAUDE.md pointer (#338)#338133 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(ci): scheduled Neon orphan-branch cleanup workflow (#337)#337122 lines
  • docs(solutions): capture React Compiler cleanup arc patterns + refresh stale gu…#339229 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • ci(lint): ratchet --max-warnings 172 (post-plan-005 baseline) (#335)#3355 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(plans): mark ci-residuals-006 done + resolution summary (#336)#33615 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • chore: gitignore worktree-manager .env.local.example.backup* files (#334)#3344 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • refactor(react-compiler): slice 9 — clear last set-state-in-effect + re-elevate…#333197 lines
  • refactor(react-compiler): slice 8 — clear 2 set-state-in-effect in VoiceConfide…#33237 lines
  • refactor(react-compiler): slice 7 — clear 3 set-state-in-effect in AgentFeedPan…#33157 lines
  • refactor(react-compiler): slice 6 — clear 2 set-state-in-effect + 1 preserve-ma…#33097 lines
  • refactor(react-compiler): slice 5 — clear purity warnings via useNow hook (#329)#329148 lines
  • refactor(react-compiler): slice 4 — clear set-state-in-effect in 3 sync-state-t…#32863 lines
  • feat(v3a-small): Pattern A pending template refresh + V3a-medium plan supersede…#326355 lines
  • fix(v2): SELECTOR_TYPE_DRIFT pending template — copy fix + decoupled sentinel t…#32567 lines
  • fix(hooks,vault): React Compiler refs + immutability cleanup — slice 3 (#323)#323100 lines
  • docs(brainstorm): V3a SELECTOR_TYPE_DRIFT Pattern B redesign — Option C forced;…#3241915 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(hooks): clear react-hooks/refs warnings in 3 polling/selection hooks (slice…#32224 lines
  • docs(solutions): capture mountedRef latch finding + refresh adjacent React-hook…#320136 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(v3a): PR3/5 — U13 chip-dispatch React components (IntentChipSet + IntentSu…#3191285 lines
  • feat(v3a): PR2/5 — Phase 2 chip-dispatch endpoint (U12 + U14) (#317)#3171488 lines
  • docs(brainstorm): capture 4 G2-α deferred design questions as G2 follow-ups (ro…#318330 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(observations): fill G2-α window timeline + open watch (post-deploy) (#316)#31623 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(v3a): PR1/5 — action-class registrations (U7+U8+U9) (#313)#313876 lines
  • feat(vault): G2-α activation — register gov.key_staff + UI rewrite + extraction…#3061560 lines
  • feat(v3a): Phase 1 foundation — chip primitive infrastructure (8/17 units) (#31…#3129959 lines
  • test(llm): introduce mockMeasured<T> helper to type-narrow measuredCall fixturesc30f4a6179 linesfolded into the change before it
  • docs(runbook): document recovery for cancelled push:main run blocking branch pr…dcaf61a8 linesfolded into the change before it
  • refactor(vault-tabs): widen getMissingAtomsForTab param to accept untrusted str…feadcfd5 linesfolded into the change before it
  • fix(workspace): extend race fixes per ce:review (setError guards, Cancel/Escape…0ed512254 lines
  • fix(workspace): harden three SelectionActionBar lifecycle races597425c36 linesfolded into the change before it
  • docs(plans): capture deferred SelectionActionBar race findings from ce:reviewa7ea7b782 linesfolded into the change before it
  • refactor(react-compiler): clean set-state-in-effect — slice 1/N (3 files) (#308)#308215 lines
  • feat(vault): G2-α capability — collapse helper + back-compat seam + runbook + o…#3055940 lines
  • docs(ci-runbook): note GitHub Pro requirement for branch protection (#307)#3078 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): closure-stamped trusted-parameter pattern + V1 timing-race doc…#304267 linesfolded into the change before it
  • feat(ci): CI/CD baseline — GitHub Actions + Neon branch per run (#303)#3031532 lines
  • feat: V2 flag-family classifier + splice action classes (#302)#3022378 lines
  • Merge pull request #301 from Drewx-Design/docs/latent-guard-audit-pattern#301163 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(agent): scope propose_passage_replacement vault lookup by programId (#299)#299557 lines
  • Merge pull request #300 from Drewx-Design/docs/refresh-fanout-discipline-and-ad…#300160 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #298 from Drewx-Design/feat/record-array-selector-grammar#2981332 lines
  • Merge pull request #297 from Drewx-Design/feat/lever-1-refactor-routing-source-…#2973192 lines
  • feat: Agent action-posture V1 — propose-and-confirm splice loop (#293)#2938916 lines
  • Merge pull request #296 from Drewx-Design/docs/g1-observation-window-empirical-…#296151 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(test): green main — eliminate failing tests (#295)#295392 lines
  • Merge pull request #294 from Drewx-Design/feat/lever-1-prompt-threading#2941588 lines
  • observability: emit LLM_CLASSIFIER_SOFT_FAILURE for evaluate-derived-metrics ca…#29241 lines
  • Merge pull request #288 from Drewx-Design/docs/path-a-prime-zod-schema-topology#288296 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #287 from Drewx-Design/feat/fu-23-pr-c-activate-fin-revenue-…#287294 lines
  • Merge pull request #286 from Drewx-Design/docs/observation-windows-fu123-merged#286189 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #280 from Drewx-Design/fix/fu-1-2-3-compound-atom-prose-leak#2801795 lines
  • feat(llm): A.1 generate-section routing seam — Opus 4.7 on day one + Path B fou…#281157 lines
  • feat(llm): retarget generate-plan::call_1 to claude-opus-4-7 + adaptive thinkin…#278442 lines
  • Merge pull request #279 from Drewx-Design/feat/fu-23-pr-c-activate-prog-logic-m…#279185 lines
  • Merge pull request #276 from Drewx-Design/docs/g1b-observation-window-pending-f…#27690 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(llm): upgrade Tier B.1 evaluators (haiku-4.5 → sonnet-4.6) (#277)#277904 lines
  • Merge pull request #275 from Drewx-Design/feat/fu-30-prog-logic-model-direct-lo…#275638 lines
  • Merge pull request #274 from Drewx-Design/feat/fu-23-pr-c-activate-org-address-…#274146 lines
  • Merge pull request #273 from Drewx-Design/docs/g1b-runbook-fu24-verification#27365 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #272 from Drewx-Design/feat/fu-24-direct-lock#2722791 linesfolded into the change before it
  • Merge pull request #271 from Drewx-Design/feat/fu-24-prompt-rule-string-type#27146 lines
  • docs(solutions): LLM-compliance dry-run discipline during plan authoring (#270)#270156 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(g1b): runbook dispatch learnings — "data" envelope + expected_ranges overr…#26959 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(g1b): Phase B halt observation log + FU-24 finding (#268)#268188 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(g1b): correct F5 mental model — validator reads formatAddress(value), not …#26746 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(vault): G1B tranche — lock org.address_primary selector_schema (#266)#2661552 lines
  • feat(vault): add "string" variant to SelectorFieldSchema (#265)#265340 lines
  • feat(fu-17): fu17_refused_writes table + schema_widen enum extension (Unit 7)01979ac3981 linesfolded into the change before it
  • refactor(fu-17): rename backfill-compound-atoms.ts → compound-schema-writers.ts…030f5b44890 linesfolded into the change before it
  • feat(fu-17): catch shells route FU-23 errors to fu17_refused_writes (Unit 9)1b3d542588 linesfolded into the change before it
  • fix(fu-17): 422 body shape + strip enumeration-probe vectors (review #8 + #16)2d088e7202 linesfolded into the change before it
  • fix(fu-17): silent-retry cascade — always-throws shell + audit-failure guard (r…31ae43e177 linesfolded into the change before it
  • docs(fu-17): land paired-sprint plan + requirements brainstorm with post-review…3a8bd6a1238 linesfolded into the change before it
  • fix(fu-17): partial unique index + ON CONFLICT dedup for refused-writes (review…43e187e3960 linesfolded into the change before it
  • fix(fu-17): widen correctness bundle — tx param, org filter, program scoping (r…9a653db519 linesfolded into the change before it
  • feat(fu-17): widenSelectorSchema with 3 guards + backward-conformance (Unit 8)9c7dc52769 linesfolded into the change before it
  • fix(fu-17): split FU23CallerSite into DbRouted / Internal unions (review #18)a1a917e77 linesfolded into the change before it
  • docs(fu-17): skills guidance for schema_violation response + fix support-channe…a329c3c30 linesfolded into the change before it
  • feat(fu-17): admin refused-writes triage view + CLAUDE.md v1b (Unit 10)c106fd53996 linesfolded into the change before it
  • fix(fu-17): admin DB access role-scoping (review #2 + #1)cdb0c753846 linesfolded into the change before it
  • refactor(fu-17): apply safe_auto fixes from /ce:reviewce9ed1e80 linesfolded into the change before it
  • fix(fu-17): three-way backward-conformance check accepts fields-being-added (re…d9075c5221 linesfolded into the change before it
  • fix(fu-17): revert verify-route document state on FU-23 422 (review #9)ed90b2645 linesfolded into the change before it
  • fix(fu-17): lookup-990 per-mapping tx + continue-variant catch shell (review #3)fe1f8f1271 linesfolded into the change before it
  • feat(fu-23): altVersions-append + demote-validation helper invocations (Unit 3)05a5df8109 lines
  • test(fu-23): ADV-7 isolated-clause S8b tests + writer-site audit doc3391448387 linesfolded into the change before it
  • feat(fu-23): spread-into-value AST lint + CLAUDE.md v1a entry (Unit 5)4103256353 linesfolded into the change before it
  • feat(fu-23): shared typed errors + conformance helper (Unit 1)53a5eeb484 linesfolded into the change before it
  • feat(fu-23): wire conformance helper at primary-value-writer sites (Unit 2)5f8b4df356 linesfolded into the change before it
  • feat(fu-23): structural guards on privileged/sentinel writers (Unit 4)687d088327 linesfolded into the change before it
  • refactor(fu-23): apply safe_auto fixes from /ce:review9cc3c4578 linesfolded into the change before it
  • style(in-review): restore cream/white tier split below chips, above Certificati…517dc9634 linesfolded into the change before it
  • fix(in-review): unpin Certification Gate so it moves with chip expansiondec86d943 lines
  • fix(left-panel): add scroll to Intake/Planning panels and Outline tab04941fc6 lines
  • docs(solutions): ce-review fan-out discipline as institutional practice (#262)#262143 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(fu-25): retroactive audit of already-compound atoms — gate artifact (#261)#2611234 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #251 from Drewx-Design/feat-in-review-component-restyle#2517365 linesfolded into the change before it
  • Merge pull request #248 from Drewx-Design/feat-in-review-visual-polish-needs-you#2484586 lines
  • docs(claude-md): add Never-Do entries for Anthropic Zod patterns + writer-site …#2608 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): G1 milestone + Option B second-writer discovery (#259)#259224 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(backfill): emit concrete per-field JSON Schema types for Anthropic extracti…#25895 lines
  • fix(backfill): extraction schema names expected fields + forensic diagnostics (…#25786 lines
  • fix(backfill): extraction schema uses z.looseObject to avoid propertyNames reje…#25652 lines
  • chore: pivot G1 to fin.revenue_breakdown single-atom canary + restructure follo…#255146 lines
  • chore(gitignore): exclude Inngest run dumps + prod-data scratch pathsa95b6cf11 lines
  • fix(backfill): remove maxItems from classifier Zod schema (Anthropic rejects) (…#25438 lines
  • feat: ref grammar selector + compound-atom migration (certification unblock) (#…#25317980 lines
  • docs(solutions): cross-reference planDirty doc to UI-lockdown playbook78a40114 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • docs(solutions): capture UI-lockdown multi-surface write-audit playbookc3aab22368 linesfolded into the change before it
  • Merge pull request #252 from Drewx-Design/refactor/workspace-header-chrome#2521302 lines
  • Merge pull request #250 from Drewx-Design/docs/solutions-artifact-vs-funder-tex…#250178 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #249 from Drewx-Design/refactor/tier-nomenclature-user-facin…#249610 lines
  • fix: board brief modal no longer renders half off-screen5f2b8792 linesfolded into the change before it
  • Merge pull request #247 from Drewx-Design/chore/relocate-orphan-docs-and-file-m…#2471011 lines
  • Merge pull request #246 from Drewx-Design/docs/session-artifacts-supplemental-2…#246862 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #245 from Drewx-Design/docs/session-artifacts-backfill-2026-…#24534340 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #244 from Drewx-Design/fix/certify-zod-strict-and-role-gate#2444 linesfolded into the change before it
  • Merge pull request #243 from Drewx-Design/refactor/remove-dormant-concierge-rev…#243849 linesfolded into the change before it
  • Merge pull request #242 from Drewx-Design/docs/refresh-solutions-batch-2026-04-…#242914 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #241 from Drewx-Design/fix/anchor-ref-hallucination-upstream#241229 lines
  • Merge pull request #240 from Drewx-Design/fix/logic-sentry-system-prompt-shape#240349 lines
  • Merge pull request #235 from Drewx-Design/feat/unit-7-analyze-llm-audit#2353276 lines
  • Merge pull request #238 from Drewx-Design/docs/compound-voice-gate-catch-pattern#238137 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #237 from Drewx-Design/docs/compound-learnings-sticky-critic…#237264 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #236 from Drewx-Design/feat/voice-authenticity-gate#2362382 lines
  • Merge pull request #234 from Drewx-Design/feat/submission-path-aware-in-review#2342579 lines
  • Merge pull request #229 from Drewx-Design/feat/unit-6-day1-reconciliation#2292868 lines
  • Merge pull request #233 from Drewx-Design/fix/5c-typed-mock-as-never-casts#23312 lines
  • Merge pull request #226 from Drewx-Design/feat/unit-5c-ci-guards-and-cleanups#2261155 lines
  • Merge pull request #232 from Drewx-Design/fix/cert-gate-tier-2-3-null-bypass#232356 lines
  • Merge pull request #231 from Drewx-Design/fix/eligibility-haiku-fallthrough-tel…#231505 lines
  • Merge pull request #230 from Drewx-Design/fix/default-tab-health#23054 lines
  • Merge pull request #228 from Drewx-Design/docs/radix-tabs-forcemount-learning#228167 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #224 from Drewx-Design/feat/in-review-navigation#22416068 lines
  • Merge pull request #227 from Drewx-Design/refactor/surface-grounding-exhaustion#227115 linesfolded into the change before it
  • Merge pull request #225 from Drewx-Design/refactor/vault-predicate-extraction#225332 lines
  • Merge pull request #220 from Drewx-Design/feat/unit-5a-migrate-call-sites#2201492 lines
  • fix(voice-preview): align with review-chat pattern — abortSignal + dual-audit c…5c909b811 lines
  • feat(llm): Unit 5b — migrate 3 streamText sites to measuredStreama729ffe64 linesfolded into the change before it
  • docs(llm): clean up comment rot from Unit 5b reviewbaae27d14 linesfolded into the change before it
  • fix(llm): Unit 4 correction — raise stream GC-gotcha timeout above maxDurationeab73e341 linesfolded into the change before it
  • Merge pull request #223 from Drewx-Design/fix/atom-grounding-quantitative-claims#2232092 lines
  • Merge pull request #222 from Drewx-Design/fix/roadmap-loader-abort-resilience#222338 lines
  • Merge pull request #219 from Drewx-Design/docs/compound-verify-before-ship-lear…#219327 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #211 from Drewx-Design/feat/cached-by-default-llm-wrapper#21117941 lines
  • Merge pull request #217 from Drewx-Design/fix/compliance-checklist-stale-addres…#2171868 lines
  • Merge pull request #216 from Drewx-Design/docs/compound-learnings-dashboard-red…#216460 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #215 from Drewx-Design/fix/voice-baseline-null-check#21527 lines
  • Merge pull request #214 from Drewx-Design/feat/dashboard-editorial-redesign#2143264 lines
  • Merge pull request #213 from Drewx-Design/docs/adversarial-only-escalation-patt…#213112 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • feat(missing): Missing<Reason> / fallback primitive — thin pilot (Option A) (#2…#2121725 lines
  • Merge pull request #210 from Drewx-Design/docs/inngest-step-timer-learning#210261 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #202 from Drewx-Design/feat/voice-quality-overhaul#2023505 lines
  • Merge pull request #204 from Drewx-Design/chore/compound-ce-review-after-audit#204243 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • Merge pull request #203 from Drewx-Design/refactor/security-audit#20317073 lines
  • docs: compound learning — copilot cross-panel state and project feed atom patte…0274d03189 linesnot counted: 5 or fewer lines, or docs only, and no one else reviewed it
  • fix(copilot): move roadmap from pinned widget into feed (#201)#201242 lines
  • feat(copilot): cross-panel score focus — left panel navigates, right panel acts…#2001917 lines
  • feat(copilot): refine roadmap — summary density, step consolidation, visual hie…#199469 lines
  • docs: compound learning — async Jotai state coordination reliability patterns8a89dc2172 linesfolded into the change before it
  • feat(copilot): progress-first panel hierarchy (#198)#198393 lines
  • fix(deploy): rename [sectionId] routes to [sectionKey] for Next.js slug consist…55295d4346 linesfolded into the change before it
  • Merge pull request #197 from Drewx-Design/feat-copilot-surface-m2#1979322 lines
  • 221 earlier changes (counted; not itemized)