FIN Weekly digest — 2026-W26
This week’s work pushed FIN’s agentic stack from guarded proving into live, gate-first operation: the biggest step was arming central Hermes execute in production while simultaneously standing up real invariant scoring, qualifying-read protections, and proof trails around it. In parallel, the team turned “heal” from a placeholder into an actual closed loop, with genuine tenant self-heal and a central content-heal path reaching real production targets. The through-line was clear: more autonomy, but only with tighter invariants, better observability, and reversible rollout mechanics.
Highlights
- Central execution crossed an important threshold with #1613, which armed
FIN_CENTRAL_HERMES_EXECUTE_ENABLEDin production in a gate-first, fail-closed, reversible way — the clearest sign this period that autonomous central actions are moving from shadow mode into controlled reality. - The safety case around that move got materially stronger through #1587, a real invariant scorer for
resolver_routing_replay, and #1589, which flippedHERMES_EVAL_GATE_MODEfrom shadow to enforce in prod, backed by proof work in #1622. - Loop closure became substantive rather than symbolic with #1618, replacing hollow tenant repair with genuine tenant knowledge self-heal, then hardening and extending that path in #1623.
- Central healing also matured quickly: #1625 reworked central heal into a true source→profile JOIN/fold, and #1626 activated central content-heal against 125 real production targets alongside tenant heal, still under gate-first controls.
- The central knowledge base gained a much firmer health contract through #1606, which introduced a real-invariant
central-kb-health-v1scorer, followed by fixes in #1608 and #1610 to land live-harvest end-to-end and get that health check to persist a pass. - Data-engine reliability work kept pace with the autonomy push: #1602 added a backlog-adaptive sharded projection-rebuild drain, and #1604 armed the production batch path with the reaper, sharded drain, and scheduled crons to close the drift loop.
Work started
| Repo | PR | Title | Author | Labels |
|---|---|---|---|---|
fin-platform/fin-agentic | fin-platform/fin-agentic#1626 | activate(prod): central content-heal (125 real targets) + tenant heal, gate-first | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1625 | feat(loop-closure): rework central heal into a genuine source→profile JOIN/fold | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1624 | activate(prod): arm tenant + central content-heal gate-first | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1623 | batch(loop-closure): tenant-heal hardening + dream tx-abort + central content-heal + tenant-USDA mirror | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1622 | docs(proof): finalize p-eval-gate-stub-read VERIFY (landing + prod gate-read flip) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1621 | fix(eval-gate): qualifying-read so insufficient-data stubs can’t withhold execute authority | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1620 | fix(dream-cycle): stop the THINK-phase tx-abort — propagate the real error (finding #73) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1619 | harden(loop-closure): now-free idempotency + content guard + provenance for tenant knowledge heal | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1618 | feat(loop-closure): genuine tenant knowledge self-heal (replace hollow repair) [INERT/flag-off] | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1617 | docs(signoff): coordinator sign-off — keystone first autonomous act (W1-S7c/W1-S7/W7) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1616 | docs(phase4): record PHASE4-EMBEDDING-GUC merge SHA + coordinator-deploy flag | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1615 | fix(phase6): move gated-suite eval-runner 15 2 → 45 1 (gate-read margin before the 02:00 central execute) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1614 | harden(knowledge): embedding-GUC session-default coupling lands robustly on any write path | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1613 | activate(prod): arm central Hermes execute (FIN_CENTRAL_HERMES_EXECUTE_ENABLED=true) — gate-first, fail-closed, reversible | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1612 | docs(proof): PHASE6 USDA match-rate AFTER evidence (1→84 distinct commodities) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1611 | docs(phase6): add PHASE6-CENTRAL-ARM-DURABILITY tracker row (PROVING) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1610 | fix(phase6): sweep central planes so central-kb-health-v1 can persist a pass | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1609 | fix(phase6): USDA catalog→report commodity-name match (singularize+alias, profile-only catalog) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1608 | fix(phase6): land central R1 live-harvest e2e + clear central-kb-health-v1 | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1607 | docs(proof): PHASE6 central-kb-health scorer — landed+deployed+live prod eval_result | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1606 | feat(eval): real-invariant scorer for central-kb-health-v1 (P6, central mirror of W1-S7c) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1605 | docs(proof): data-engine arm batch — live reaper + sharded verification + ARMED-ON-PROD tracker | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1604 | activate(prod) data-engine arm batch: reaper + sharded-drain + 4 W2-B crons; close drift loop | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1603 | docs(tracker): PHASE2-SHARDED-DRAIN landed+deployed bookkeeping | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1602 | PHASE2(sharded-drain): backlog-adaptive sharded-parallel projection-rebuild drain (flag-off-safe) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1601 | PHASE4(adaptive-reembed): self-scaling depth+latency batch, hard-clamped, policy-in-knowledge | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1600 | PHASE4(KB): freshness scheduler + brief re-render (reuse refreshKnowledgeFreshness; flag-off-safe) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1599 | PHASE3(resolver): wire resolver.reviewed consumer — relearn/demote alias | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1598 | docs(phase2): idle-tx reaper deploy + staging-soak evidence | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1597 | PHASE2(drain-telemetry): health.drain_tick + deterministic drain-health classifier | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1596 | PHASE2(drain-canary): idle-in-transaction reaper — staging-armed / prod-inert | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1595 | feat(PHASE0): process-registry drift detector — monitored process-map invariant | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1594 | feat(PHASE0): assert the mobile-QC closed loop e2e — wire correction→relearn + loop-liveness canonical | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1593 | feat(PHASE0): level-1-task-runner fail-closed liveness — ops.level1.runner_no_tasks on a zero-due tick | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1592 | feat(PHASE0): capture.attempt_received completeness SLI — stop silent webhook loss | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1591 | hardening(PHASE0): embedding-GUC fail-loud assert at the single chokepoint | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1590 | fix(W7): operator reads latest QUALIFYING confirm-in-window, not single-latest row | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1589 | activate(W1-S7): flip prod HERMES_EVAL_GATE_MODE shadow→enforce (owner-approved) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1588 | docs(W1-S7c): proof bundle + tracker PROVING for resolver_routing_replay scorer | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1587 | W1-S7c: real invariant scorer for resolver_routing_replay (keystone, flag-off-safe) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1586 | W8-PKB-2: wire fin-meeting-prep to the personal-knowledge producer (flag-OFF inert) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1585 | docs(W4-G2): post-deploy USDA fetch-fix proof evidence + tracker SHA | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1584 | fix(W4-G2): USDA refresh date-window lookback + fetch-diagnostics visibility | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1583 | W8-PKB: per-user personal-knowledge skill-observation producer (flag-OFF) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1582 | fix(llm-router): classify transient Workers network errors as retryable network_error (W6-E) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1581 | docs(W2): coordinator sign-off W2-E1/E2/E3 + measure-sweep reconciliation | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1580 | fix(W2-E3): synth embedding-model GUC + renderer source-fetch RLS transaction (cron self-heal) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1579 | feat(W2-E2/E3): projection-aware vendor/customer renderers + 360-synthesis (operational living-wiki 0→N) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1578 | docs(W2-E1): after-evidence — operational lot doc rows 0→203 on prod | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1577 | feat(W2-E1): operational living-wiki doc rows — seeder + renderer handle operational entities | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1576 | feat(W3-G3): recurring retrieval-eval prod cron emitting the real P@5 scorecard | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1575 | chore(W4-G2): bind USDA_MARKET_NEWS_API_KEY to fin-agentic-prod + proof bundle | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1574 | fix(dream-cycle): GUC-independent embedding_model on Sweep D derived UPSERTs (W6-D loop-unblock) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1573 | docs(tracker): reconcile master proof-tracker to the measured-reality re-audit (2026-06-27) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1572 | docs(W2): sign off resolver keystone — W2-C3 + W2-C4 SIGNED-OFF (link-rate 99.77%) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1571 | feat(central-hermes): FIN_CENTRAL_HERMES_EXECUTE_ENABLED in wrangler (default OFF) + prove execute e2e wiring (W1-S3, Lock 2) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1570 | feat(W2-C3): synonym/variety map + high-confidence master-root collapse clears the resolver fuzzy tail | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1569 | docs(W2): coordinator sign-off W2-C0/D1/D2 + re-route D3/D5 + correct stale resolver gap | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1568 | docs(proof): W2-C1 coordinator sign-off + W2-D1/D2/D3/D5 gap diagnosis | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1567 | docs(W2-C0): prove commodity master/alias index keystone (link-rate 0.09% → 66.7%) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1566 | feat(operator): engine-control tools — operator triggers dream-cycle/eval/mirror/re-embed (W1-S2, flag OFF) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1565 | feat(resolver): pure commodity-word singularizer for fuzzy-tier root collapse | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1564 | perf(resolver): trim candidate-load content payload to unblock projection-rebuild sales wall (#14) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1563 | docs(program): FIN master-program proof tracker + proof-bundle scaffold | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1562 | fix(rebuild): transaction-local lock_timeout to unblock projection-rebuild throughput (#23) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1561 | docs(program): FIN agentic-intelligence master execution program (W0–W8) from 4-audit synthesis | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1560 | perf(rebuild): PROJECTION_REBUILD_CURSOR_FLUSH_EVERY=1 — stop the drain parking on dense regions | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1559 | fix(resolver): enforce corroboration>=2 at the execution-authorization seam + earn it from genuine historical multi-source agreement | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1558 | feat(rebuild): thread cursorFlushEvery through the off-Worker rebuild CLI | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1557 | fix(resolver): emit honest verdict_basis on auto-link/approve confirm rows | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1556 | perf(rebuild): kill the 5-min system-health 1.5M-row full-tenant scan contending with the rebuild | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1555 | fix(operator): normalize ops-backlog datetimes to strict ISO — revive the dark daily operator (S6b) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1554 | fix(rebuild): self-reap the orphaned session advisory lock — make the drain continuous | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1553 | fix(cqrs): unblock the projection-rebuild drain — idle-tx reaper + incremental cursor + re-enable queue | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1552 | feat(central-hermes): RLS-scoped operator EXECUTE role + execute-path wiring (S5) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1551 | revert(cqrs): disable queue-rebuild — import zombie-txn leak is the deeper blocker | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1550 | feat(central-hermes): genuine LLM-reasoning turn reusing the tenant reasoning core (S4, flag-OFF byte-identical) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1549 | fix(cqrs): effective statement_timeout through Hyperdrive — unblock the projection-rebuild drain + re-enable queue | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1548 | revert(cqrs): disable queue-rebuild drain — protect live import path | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1547 | fix(hermes-operator): isolate eval-gate read from tenant emit tx — revive dark daily operator (S6) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1546 | feat(skill-invoker): bind real structuring tool to act-layer apply hook (authority-gated, inert under shadow) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1545 | docs(central-hermes): genuine Central Hermes agent + operator-as-engine-MANAGER spec (owner directive 2026-06-26) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1544 | feat(cqrs): dedicated projection-rebuild queue — stop tick starvation behind PP-CSV imports | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1543 | feat(digest): living editorial GitHub-activity digest → FIN Central intranet | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1542 | feat(producesim): score REAL tenant predictions vs REAL outcomes (all families) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1541 | chore(cqrs): enable queue-driven projection-rebuild durable drain on prod | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1540 | feat(cqrs): queue-driven projection-rebuild — CF-native durable drain (flag-gated, NO-OP off) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1539 | feat(wave-e): vendor living-wiki doc from vendor_360 — first operational doc-graph + backlinks | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1538 | feat(retrieval): activate FIN_RETRIEVAL_TEMPORAL_ENABLED on prod (gbrain task #58) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1537 | fix(reembed): durable INSERT-path enqueue trigger for NULL-embedding knowledge rows (mig 101) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1536 | fix(resolver): thread named 6-tier tier_name into resolver.* event metadata (#57) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1535 | fix(po-timeline): twin-lot coverage gap — prod still double-counted received_cases after #1531 | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1534 | feat(projection-rebuild): backlog-starvation observability + commodity_360 revenue invariants | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1533 | feat(eval): point BrainBench at the REAL retrieval adapter (not the dumb lexical baseline) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1532 | feat(retrieval): temporal-ordering capability for query_knowledge (gbrain task #58) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1531 | fix(po-timeline): kill live received/ordered double-count (2x on prod) + fill-rate/short-ship | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1530 | fix(llm): kill the structured_json envelope-mismatch bug class (raw_json for domain-parse sites) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1529 | feat(cron): weekly commodity self-wiring refresh — keep the RRF graph channel’s typed edges fresh | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1528 | fix(think-op): raw_json synthesis + activate FIN_THINK_OP_ENABLED (dormant-flag re-grade) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1527 | P6 Gate-1: close the tenant→central feedback loop (proven on staging) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1526 | fix(lsd): contradiction-judge uses raw_json — root-cause of 0 LSD proposals on prod | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1525 | feat(grain-bridge): Wave D commodity_360 grain-bridge + revenue/margin (+migration 100) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1524 | feat(lot-lifecycle): Wave D unit_cost/sold pricing/received_date columns (+migration 099) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1523 | fix(graph): scope async doc-backlink rewrite away from self-wiring rows (P1 follow-up to #1522) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1522 | feat(graph): self-wire the main doc write path → densify the typed-edge graph | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1521 | fix(eval): gated-suite runner records a disabled receipt so every tick is observable | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1520 | fix(dream-cycle): bind ENRICH-phase tags as real text[] (2nd text[] array-literal bug class) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1519 | feat(vendor-360): Wave D cost-basis/spend/fill-rate columns (+migration 098) — single-anchor twin-emit fix | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1518 | fix(reranker): batch + pad-partial the CF-Llama judge so it stops fail-opening ~19% of queries | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1517 | feat(cron): recurring resolver re-resolve self-heal CRON (Wave C/F loop closure) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1516 | fix(deploy): minify production release script | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1515 | fix(self-heal/A3): PDF arrival backlog-replay cron = PURE PRODUCER (queue, not inline synthesis) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1514 | fix(qc): render product master evidence in inspection reports | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1513 | fix(self-heal/A3): make PDF arrival backlog-replay cron a PURE PRODUCER (queue, not inline synthesis) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1512 | test(qc): prove social photo branch and vision paths | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1511 | perf+fix(self-heal): cut worker startup-CPU (10021) + revert PDF replay limit 5→1 | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1510 | fix(projection/inventory): cap work-thread event source_ids so heavy days clear the 64KB guard | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1509 | fix(self-heal/A3): per-orphan report-type-aware deferFanout — drain the PDF arrival backlog | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1508 | fix(ops/freshness): move heavy central-read lanes off the saturated 0 2 cron to offset minutes | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1507 | fix(self-heal/A3): wrap PDF backlog-replay synthesis in connection-reuse frame (orphans finally drain) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1506 | fix(projections/Wave-D): light up commodity_360 velocity + customer_360 orders-grain from sales.report_ingested | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1505 | fix(self-heal/A3): window arrival_report backlog-replay so orphans drain | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1504 | feat(resolver/Wave-C): commodity_master index — unblock commodity entity-linking | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1503 | feat(engine): Wave A — de-risk + one-line unblocks (eval P0, 10021 startup-CPU, PDF canary, empirical guards) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1502 | fix(projections): customer_360 + daily_pulse revenue from sales.report_ingested — closes the sales→revenue loop | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1501 | fix(import): suppress per-event fan-out for bulk sales imports — the real SALES LOG DAILY OOM fix (slice-2) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1500 | fix(import): sales-aware chunk slice — stop the SALES LOG DAILY OOM (sales.report_ingested=0) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1499 | fix(arrivals): chunk per-arrival W8.12 synthesis — unwedge the OOM-stalled shared drops consumer (#11) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1498 | fix(import): chunk large Produce Pro CSV imports across queue invocations (completes the 8086-row memory fix) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1497 | config(import): activate per-row fan-out deferral on prod (#1495 activation — fixes large-feed timeout) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1496 | fix: clock-freeze cross-interface-memory test + observable PDF backlog-replay telemetry | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1495 | fix(import): batch sales-feed inserts + defer per-row fan-out — kill the 8086-row import timeout | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1494 | fix(reranker): make the CF-Llama cross-encoder judge actually parse (Workers AI returns pre-parsed JSON) + size its token budget | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1493 | fix(eval): #37 unblock full retrieval suite past cold vector-scan timeout | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1492 | fix(ingestion): register ingest.raw_email_capture* event types (capture telemetry was dropped in prod) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1491 | feat(ingestion): durable raw-inbound capture — never drop arrivals@/intel@ reports on failure | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1490 | fix(ingestion): classify “Confirmation of Sale” PDFs → unblock sales feed → customer_360 | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1489 | fix(cron): PDF backlog-replay limit 5→1 (autonomous synthesis completes within budget) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1488 | fix(memory): thread now through getRecentContext (unblocks the gauntlet — latent retention time-bomb) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1487 | fix(cron): PDF backlog-replay dedicated cron budget + brain-visible telemetry (F1 — the QC-queue fix) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1486 | fix(operational-spine): gated-suite eval-runner per-run receipt + eif nightly datetime coercion | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1485 | fix(qc-history): degrade gracefully when Teams narrative table is absent (non-Teams population path) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1484 | fix(commodity-mirror): GUC-independent embedding metadata on tenant shared upsert (spine freshness) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1483 | fix(ingest): CSV backlog-replay newest-first (unblocks the starved po_received receipt feed) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1482 | config(reranker): make CF-Llama cheap lane durable in wrangler.jsonc (prod+staging) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1481 | fix(ingest): CSV replay uses subject-authoritative stored report_type (po_received receipt feed) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1480 | fix(ingest): time-bound PDF backlog-replay so the self-heal completes within budget | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1479 | fix(ingest): FIN Phase 0 — durable arrivals@ PDF precapture (Step 5) + backlog-replay self-heal | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1478 | fix(ingest): RP581 arrival-PDF queue-loss fix + pricing + intel@ department fallback | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1477 | fix(ingest): RP581 PDF authoritative QC-queue source + subject-driven classification + provisional fallback + 6b decouple | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1476 | fix(projections): derive commodity for supply_chain_position via canonicalization + po_timeline bridge | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1475 | feat(qc-history): wire the non-Teams population path for qc_inspection_history_projection | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1474 | fix(dream-cycle): bind think-op tags as real text[] + isolate receipt emit in a savepoint | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1473 | fix(po-timeline): derive canonical po_id from po_number so arrival/receiving/QC evidence spines | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1472 | chore(projections): retire 3 vestigial CQRS phantoms (op-record-search, knowledge-context-index, qc-work-queue) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1471 | fix(projections): render s26_competitive + chain doc_backlink backfill on activation | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1470 | fix(migration): thread DB timeout env knobs through buildMigrationEnv | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1469 | feat(retrieval+graph): commodity-aware lot-identity ordering + operational doc-graph edge-builder (P1) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1468 | feat: graph-edges from linked_entities + BrainBench sealed corpus | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1467 | feat(reranker): rewire cross-encoder judge to CF US open-weight model | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1466 | chore(prod): activate cross-encoder reranker on production (FIN_CROSS_ENCODER_RERANK_ENABLED=true) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1465 | feat(qc): harden social photo post trust states | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1464 | docs: record QC production sync proof | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1463 | QC social-photo inspection E2E proof | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1462 | feat(retrieval): real LLM-judge cross-encoder scorer + wire it so rerank reorders (gbrain-parity P1-L2) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1461 | feat(prod-brain): entity_doc_projection seeder (operational, no-execute) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1460 | fix(p0): reembed worker — pre-filter non-embeddable input + per-item batch isolation | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1459 | feat: Wave-2 batch — weather mirror + tenant→central feedback + skill breadth batch 2 | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1458 | feat(eval): close 3 of 4 gated-suite canonicals with real-event invariants | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1457 | fix(dream-cycle): per-phase savepoint isolation so one phase throw can’t zero the pass (prod revival) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1456 | fix(cron): consolidated cron-fixes batch — commodity-mirror Worker-conn + eval-runner emit-resilience + Worker observability | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1455 | feat(p2+p5): Wave-1b clean batch — think-op wrapper + corpus-env hygiene + skill-breadth + mode-bundles | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1454 | feat(p2+p4+p5): Wave-1 depth — python eval bridge + routing resolver + schema-pack authority + freshness SLA | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1453 | activate(prod): enable commodity-profile mirror (FIN_COMMODITY_MIRROR_ENABLED) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1452 | feat(p4): central→tenant commodity-profile mirror (the correct two-plane freshness fix) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1451 | activate(prod): validation activation — max-pool + dream-enrich + LSD-apply + gated-suite-eval-runner | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1450 | feat(p5): autonomous enrichment — dream-cycle enrich phase + think/synthesis op | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1449 | feat(p4): bounded eval-gated Central-Hermes execute authority (owner-granted, inert) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1448 | Release QC social photo inspection and Nathel auth proof | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1447 | feat(p5): derived-enrichment substrate + immutability fence | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1446 | feat(p2+p3): eval-gate robustness (judge fingerprint + anti-gaming) + operator verdict reader | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1445 | feat(p2): enforce-readiness — eval-gate suite-id resolution + gated-suite eval-result emitters | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1444 | fix(p0): token-safe embedding chunker — never exceed the 8192-token model limit | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1443 | feat(p3): close the loops — LSD-supersession applier + tenant-intel producer/consumer + operator authorization wiring | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1442 | feat(p3): close-the-loop prep — deriveExecutionAuthorization + resolveConfig live + mission-packet pre-gate + knobs_hash cache-key | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1441 | feat(p1+p2): max-pool embedding + cross-encoder reranker + query-anchor graph seed + canonical eval suite-id registry | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1440 | feat(p0): prod-brain data-plane operational scripts (apply-065/066 runbook, backlink backfill, reembed bulk-drain, coverage verifier, graph probe) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1439 | feat(mcp): register find_trajectory + wire it as the cheap-signal gate before the LSD judge | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1438 | feat(activation): graph-channel + schema-pack (both envs) + eval-gate shadow (prod) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1437 | fix(retrieval): fail-open guard on graph RRF channel (42P01 → vector+lexical) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1436 | activate(prod): R1 live-harvest daily cron (un-starves the ingestion/graph cluster) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1435 | P2 retrieval: deterministic query expansion (synonym/alias) + activate | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1434 | P5 LSD: wire contradiction-judge into the dream-cycle + activate (+ R1 cron guard) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1433 | Parity wirings: r1-harvest cron-guard + LSD dream-cycle caller + find_trajectory tool + activation | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1432 | feat(skill-invoker): real execution recipe — propose→act genuinely closes | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1431 | activate(prod) Tier 2: structuring/serving/eval engines (11 audited-safe flags) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1430 | activate(prod) Tier 1: read-path retrieval/routing channels (owner-approved) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1429 | Activate eval-gate SHADOW on staging (close the EVAL/HEAL seam) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1428 | P3: composition_skill producer — complete the mentor loop end-to-end | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1427 | P3: mentor Evidence-Pack producer wiring (shadow, flag-gated) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1426 | P6: tenant→central intel proposal emitter (proposal-only, flag-gated) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1425 | test(qc): expand frontier state lab and repair Nathel role proof | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1424 | P3+P5: mentor response-composition + LSD trio (BudgetTracker, find_trajectory, contradiction-judge) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1423 | activate(staging): enable landed engine flags (central Hermes, skill-invoker, self-wiring, dream-cycle, source-registry) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1422 | Next-wave batch: tenant skill-invoker + mentor Evidence Pack + self-wiring graph | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1421 | feat(qc): refine social photo inspection composer | robertolouisgarcia-maker | — |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#34 | fix(contracts): sync GDF sha256 to governed hash | robertolouisgarcia-maker | — |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#33 | fix(contracts): sync GDF schema to governed fin-agentic copy (cross-repo drift) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1420 | P1-P4 batch: Central Hermes + pure-mirror + vendor-label gate + nDCG eval + retrieval modularize | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1419 | fix(p0): re-seed eval-gate personas in staging deploy — unblock the activation backlog | robertolouisgarcia-maker | — |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#32 | feat(c1): wire the raw-lake → central-corpus structuring loop (the root unlock) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1418 | docs(program): FIN Agentic ERP Execution Program (2026-06-22) — the framed plan | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1417 | Fix QC simulator auth proof harness | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1416 | fix(cron): un-strand heartbeat + QC wiki-writer crons in production (flag-gated) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1415 | fix(sse): flush initial heartbeat before cold-start tip query — unblocks staging deploy gate | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1414 | enable(retrieval): named-thing channel ON for staging (measured vendor-resolution 4/4) | robertolouisgarcia-maker | — |
fin-platform/fin-agentic | fin-platform/fin-agentic#1413 | Retrieval: named-thing channel — vendor-resolution 0/4→4/4, overall P@5 0.27→0.45 (MEASURED) | robertolouisgarcia-maker | — |
Work completed
Merged pull requests
| Repo | PR | Title | Author |
|---|---|---|---|
fin-platform/fin-agentic | fin-platform/fin-agentic#1626 | activate(prod): central content-heal (125 real targets) + tenant heal, gate-first | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1625 | feat(loop-closure): rework central heal into a genuine source→profile JOIN/fold | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1623 | batch(loop-closure): tenant-heal hardening + dream tx-abort + central content-heal + tenant-USDA mirror | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1622 | docs(proof): finalize p-eval-gate-stub-read VERIFY (landing + prod gate-read flip) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1621 | fix(eval-gate): qualifying-read so insufficient-data stubs can’t withhold execute authority | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1618 | feat(loop-closure): genuine tenant knowledge self-heal (replace hollow repair) [INERT/flag-off] | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1617 | docs(signoff): coordinator sign-off — keystone first autonomous act (W1-S7c/W1-S7/W7) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1616 | docs(phase4): record PHASE4-EMBEDDING-GUC merge SHA + coordinator-deploy flag | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1615 | fix(phase6): move gated-suite eval-runner 15 2 → 45 1 (gate-read margin before the 02:00 central execute) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1614 | harden(knowledge): embedding-GUC session-default coupling lands robustly on any write path | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1613 | activate(prod): arm central Hermes execute (FIN_CENTRAL_HERMES_EXECUTE_ENABLED=true) — gate-first, fail-closed, reversible | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1612 | docs(proof): PHASE6 USDA match-rate AFTER evidence (1→84 distinct commodities) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1611 | docs(phase6): add PHASE6-CENTRAL-ARM-DURABILITY tracker row (PROVING) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1610 | fix(phase6): sweep central planes so central-kb-health-v1 can persist a pass | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1609 | fix(phase6): USDA catalog→report commodity-name match (singularize+alias, profile-only catalog) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1608 | fix(phase6): land central R1 live-harvest e2e + clear central-kb-health-v1 | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1607 | docs(proof): PHASE6 central-kb-health scorer — landed+deployed+live prod eval_result | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1606 | feat(eval): real-invariant scorer for central-kb-health-v1 (P6, central mirror of W1-S7c) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1605 | docs(proof): data-engine arm batch — live reaper + sharded verification + ARMED-ON-PROD tracker | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1604 | activate(prod) data-engine arm batch: reaper + sharded-drain + 4 W2-B crons; close drift loop | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1603 | docs(tracker): PHASE2-SHARDED-DRAIN landed+deployed bookkeeping | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1602 | PHASE2(sharded-drain): backlog-adaptive sharded-parallel projection-rebuild drain (flag-off-safe) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1601 | PHASE4(adaptive-reembed): self-scaling depth+latency batch, hard-clamped, policy-in-knowledge | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1600 | PHASE4(KB): freshness scheduler + brief re-render (reuse refreshKnowledgeFreshness; flag-off-safe) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1599 | PHASE3(resolver): wire resolver.reviewed consumer — relearn/demote alias | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1598 | docs(phase2): idle-tx reaper deploy + staging-soak evidence | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1597 | PHASE2(drain-telemetry): health.drain_tick + deterministic drain-health classifier | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1596 | PHASE2(drain-canary): idle-in-transaction reaper — staging-armed / prod-inert | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1595 | feat(PHASE0): process-registry drift detector — monitored process-map invariant | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1594 | feat(PHASE0): assert the mobile-QC closed loop e2e — wire correction→relearn + loop-liveness canonical | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1593 | feat(PHASE0): level-1-task-runner fail-closed liveness — ops.level1.runner_no_tasks on a zero-due tick | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1592 | feat(PHASE0): capture.attempt_received completeness SLI — stop silent webhook loss | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1591 | hardening(PHASE0): embedding-GUC fail-loud assert at the single chokepoint | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1590 | fix(W7): operator reads latest QUALIFYING confirm-in-window, not single-latest row | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1589 | activate(W1-S7): flip prod HERMES_EVAL_GATE_MODE shadow→enforce (owner-approved) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1588 | docs(W1-S7c): proof bundle + tracker PROVING for resolver_routing_replay scorer | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1587 | W1-S7c: real invariant scorer for resolver_routing_replay (keystone, flag-off-safe) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1586 | W8-PKB-2: wire fin-meeting-prep to the personal-knowledge producer (flag-OFF inert) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1585 | docs(W4-G2): post-deploy USDA fetch-fix proof evidence + tracker SHA | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1584 | fix(W4-G2): USDA refresh date-window lookback + fetch-diagnostics visibility | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1583 | W8-PKB: per-user personal-knowledge skill-observation producer (flag-OFF) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1582 | fix(llm-router): classify transient Workers network errors as retryable network_error (W6-E) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1581 | docs(W2): coordinator sign-off W2-E1/E2/E3 + measure-sweep reconciliation | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1580 | fix(W2-E3): synth embedding-model GUC + renderer source-fetch RLS transaction (cron self-heal) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1579 | feat(W2-E2/E3): projection-aware vendor/customer renderers + 360-synthesis (operational living-wiki 0→N) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1578 | docs(W2-E1): after-evidence — operational lot doc rows 0→203 on prod | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1577 | feat(W2-E1): operational living-wiki doc rows — seeder + renderer handle operational entities | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1576 | feat(W3-G3): recurring retrieval-eval prod cron emitting the real P@5 scorecard | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1575 | chore(W4-G2): bind USDA_MARKET_NEWS_API_KEY to fin-agentic-prod + proof bundle | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1574 | fix(dream-cycle): GUC-independent embedding_model on Sweep D derived UPSERTs (W6-D loop-unblock) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1573 | docs(tracker): reconcile master proof-tracker to the measured-reality re-audit (2026-06-27) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1572 | docs(W2): sign off resolver keystone — W2-C3 + W2-C4 SIGNED-OFF (link-rate 99.77%) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1571 | feat(central-hermes): FIN_CENTRAL_HERMES_EXECUTE_ENABLED in wrangler (default OFF) + prove execute e2e wiring (W1-S3, Lock 2) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1570 | feat(W2-C3): synonym/variety map + high-confidence master-root collapse clears the resolver fuzzy tail | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1569 | docs(W2): coordinator sign-off W2-C0/D1/D2 + re-route D3/D5 + correct stale resolver gap | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1568 | docs(proof): W2-C1 coordinator sign-off + W2-D1/D2/D3/D5 gap diagnosis | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1567 | docs(W2-C0): prove commodity master/alias index keystone (link-rate 0.09% → 66.7%) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1566 | feat(operator): engine-control tools — operator triggers dream-cycle/eval/mirror/re-embed (W1-S2, flag OFF) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1565 | feat(resolver): pure commodity-word singularizer for fuzzy-tier root collapse | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1564 | perf(resolver): trim candidate-load content payload to unblock projection-rebuild sales wall (#14) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1563 | docs(program): FIN master-program proof tracker + proof-bundle scaffold | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1562 | fix(rebuild): transaction-local lock_timeout to unblock projection-rebuild throughput (#23) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1561 | docs(program): FIN agentic-intelligence master execution program (W0–W8) from 4-audit synthesis | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1560 | perf(rebuild): PROJECTION_REBUILD_CURSOR_FLUSH_EVERY=1 — stop the drain parking on dense regions | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1559 | fix(resolver): enforce corroboration>=2 at the execution-authorization seam + earn it from genuine historical multi-source agreement | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1558 | feat(rebuild): thread cursorFlushEvery through the off-Worker rebuild CLI | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1557 | fix(resolver): emit honest verdict_basis on auto-link/approve confirm rows | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1556 | perf(rebuild): kill the 5-min system-health 1.5M-row full-tenant scan contending with the rebuild | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1555 | fix(operator): normalize ops-backlog datetimes to strict ISO — revive the dark daily operator (S6b) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1554 | fix(rebuild): self-reap the orphaned session advisory lock — make the drain continuous | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1553 | fix(cqrs): unblock the projection-rebuild drain — idle-tx reaper + incremental cursor + re-enable queue | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1552 | feat(central-hermes): RLS-scoped operator EXECUTE role + execute-path wiring (S5) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1551 | revert(cqrs): disable queue-rebuild — import zombie-txn leak is the deeper blocker | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1550 | feat(central-hermes): genuine LLM-reasoning turn reusing the tenant reasoning core (S4, flag-OFF byte-identical) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1549 | fix(cqrs): effective statement_timeout through Hyperdrive — unblock the projection-rebuild drain + re-enable queue | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1548 | revert(cqrs): disable queue-rebuild drain — protect live import path | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1547 | fix(hermes-operator): isolate eval-gate read from tenant emit tx — revive dark daily operator (S6) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1546 | feat(skill-invoker): bind real structuring tool to act-layer apply hook (authority-gated, inert under shadow) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1545 | docs(central-hermes): genuine Central Hermes agent + operator-as-engine-MANAGER spec (owner directive 2026-06-26) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1544 | feat(cqrs): dedicated projection-rebuild queue — stop tick starvation behind PP-CSV imports | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1543 | feat(digest): living editorial GitHub-activity digest → FIN Central intranet | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1542 | feat(producesim): score REAL tenant predictions vs REAL outcomes (all families) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1541 | chore(cqrs): enable queue-driven projection-rebuild durable drain on prod | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1540 | feat(cqrs): queue-driven projection-rebuild — CF-native durable drain (flag-gated, NO-OP off) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1539 | feat(wave-e): vendor living-wiki doc from vendor_360 — first operational doc-graph + backlinks | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1538 | feat(retrieval): activate FIN_RETRIEVAL_TEMPORAL_ENABLED on prod (gbrain task #58) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1537 | fix(reembed): durable INSERT-path enqueue trigger for NULL-embedding knowledge rows (mig 101) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1536 | fix(resolver): thread named 6-tier tier_name into resolver.* event metadata (#57) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1535 | fix(po-timeline): twin-lot coverage gap — prod still double-counted received_cases after #1531 | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1534 | feat(projection-rebuild): backlog-starvation observability + commodity_360 revenue invariants | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1533 | feat(eval): point BrainBench at the REAL retrieval adapter (not the dumb lexical baseline) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1532 | feat(retrieval): temporal-ordering capability for query_knowledge (gbrain task #58) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1531 | fix(po-timeline): kill live received/ordered double-count (2x on prod) + fill-rate/short-ship | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1530 | fix(llm): kill the structured_json envelope-mismatch bug class (raw_json for domain-parse sites) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1529 | feat(cron): weekly commodity self-wiring refresh — keep the RRF graph channel’s typed edges fresh | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1528 | fix(think-op): raw_json synthesis + activate FIN_THINK_OP_ENABLED (dormant-flag re-grade) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1527 | P6 Gate-1: close the tenant→central feedback loop (proven on staging) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1526 | fix(lsd): contradiction-judge uses raw_json — root-cause of 0 LSD proposals on prod | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1525 | feat(grain-bridge): Wave D commodity_360 grain-bridge + revenue/margin (+migration 100) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1524 | feat(lot-lifecycle): Wave D unit_cost/sold pricing/received_date columns (+migration 099) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1523 | fix(graph): scope async doc-backlink rewrite away from self-wiring rows (P1 follow-up to #1522) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1522 | feat(graph): self-wire the main doc write path → densify the typed-edge graph | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1521 | fix(eval): gated-suite runner records a disabled receipt so every tick is observable | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1520 | fix(dream-cycle): bind ENRICH-phase tags as real text[] (2nd text[] array-literal bug class) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1519 | feat(vendor-360): Wave D cost-basis/spend/fill-rate columns (+migration 098) — single-anchor twin-emit fix | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1518 | fix(reranker): batch + pad-partial the CF-Llama judge so it stops fail-opening ~19% of queries | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1517 | feat(cron): recurring resolver re-resolve self-heal CRON (Wave C/F loop closure) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1516 | fix(deploy): minify production release script | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1515 | fix(self-heal/A3): PDF arrival backlog-replay cron = PURE PRODUCER (queue, not inline synthesis) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1514 | fix(qc): render product master evidence in inspection reports | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1512 | test(qc): prove social photo branch and vision paths | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1511 | perf+fix(self-heal): cut worker startup-CPU (10021) + revert PDF replay limit 5→1 | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1510 | fix(projection/inventory): cap work-thread event source_ids so heavy days clear the 64KB guard | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1509 | fix(self-heal/A3): per-orphan report-type-aware deferFanout — drain the PDF arrival backlog | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1508 | fix(ops/freshness): move heavy central-read lanes off the saturated 0 2 cron to offset minutes | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1507 | fix(self-heal/A3): wrap PDF backlog-replay synthesis in connection-reuse frame (orphans finally drain) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1506 | fix(projections/Wave-D): light up commodity_360 velocity + customer_360 orders-grain from sales.report_ingested | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1505 | fix(self-heal/A3): window arrival_report backlog-replay so orphans drain | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1504 | feat(resolver/Wave-C): commodity_master index — unblock commodity entity-linking | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1503 | feat(engine): Wave A — de-risk + one-line unblocks (eval P0, 10021 startup-CPU, PDF canary, empirical guards) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1502 | fix(projections): customer_360 + daily_pulse revenue from sales.report_ingested — closes the sales→revenue loop | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1501 | fix(import): suppress per-event fan-out for bulk sales imports — the real SALES LOG DAILY OOM fix (slice-2) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1500 | fix(import): sales-aware chunk slice — stop the SALES LOG DAILY OOM (sales.report_ingested=0) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1499 | fix(arrivals): chunk per-arrival W8.12 synthesis — unwedge the OOM-stalled shared drops consumer (#11) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1498 | fix(import): chunk large Produce Pro CSV imports across queue invocations (completes the 8086-row memory fix) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1497 | config(import): activate per-row fan-out deferral on prod (#1495 activation — fixes large-feed timeout) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1496 | fix: clock-freeze cross-interface-memory test + observable PDF backlog-replay telemetry | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1495 | fix(import): batch sales-feed inserts + defer per-row fan-out — kill the 8086-row import timeout | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1494 | fix(reranker): make the CF-Llama cross-encoder judge actually parse (Workers AI returns pre-parsed JSON) + size its token budget | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1493 | fix(eval): #37 unblock full retrieval suite past cold vector-scan timeout | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1492 | fix(ingestion): register ingest.raw_email_capture* event types (capture telemetry was dropped in prod) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1491 | feat(ingestion): durable raw-inbound capture — never drop arrivals@/intel@ reports on failure | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1490 | fix(ingestion): classify “Confirmation of Sale” PDFs → unblock sales feed → customer_360 | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1489 | fix(cron): PDF backlog-replay limit 5→1 (autonomous synthesis completes within budget) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1488 | fix(memory): thread now through getRecentContext (unblocks the gauntlet — latent retention time-bomb) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1487 | fix(cron): PDF backlog-replay dedicated cron budget + brain-visible telemetry (F1 — the QC-queue fix) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1486 | fix(operational-spine): gated-suite eval-runner per-run receipt + eif nightly datetime coercion | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1485 | fix(qc-history): degrade gracefully when Teams narrative table is absent (non-Teams population path) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1484 | fix(commodity-mirror): GUC-independent embedding metadata on tenant shared upsert (spine freshness) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1483 | fix(ingest): CSV backlog-replay newest-first (unblocks the starved po_received receipt feed) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1482 | config(reranker): make CF-Llama cheap lane durable in wrangler.jsonc (prod+staging) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1481 | fix(ingest): CSV replay uses subject-authoritative stored report_type (po_received receipt feed) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1480 | fix(ingest): time-bound PDF backlog-replay so the self-heal completes within budget | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1479 | fix(ingest): FIN Phase 0 — durable arrivals@ PDF precapture (Step 5) + backlog-replay self-heal | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1478 | fix(ingest): RP581 arrival-PDF queue-loss fix + pricing + intel@ department fallback | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1477 | fix(ingest): RP581 PDF authoritative QC-queue source + subject-driven classification + provisional fallback + 6b decouple | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1476 | fix(projections): derive commodity for supply_chain_position via canonicalization + po_timeline bridge | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1475 | feat(qc-history): wire the non-Teams population path for qc_inspection_history_projection | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1474 | fix(dream-cycle): bind think-op tags as real text[] + isolate receipt emit in a savepoint | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1473 | fix(po-timeline): derive canonical po_id from po_number so arrival/receiving/QC evidence spines | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1472 | chore(projections): retire 3 vestigial CQRS phantoms (op-record-search, knowledge-context-index, qc-work-queue) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1471 | fix(projections): render s26_competitive + chain doc_backlink backfill on activation | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1470 | fix(migration): thread DB timeout env knobs through buildMigrationEnv | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1469 | feat(retrieval+graph): commodity-aware lot-identity ordering + operational doc-graph edge-builder (P1) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1468 | feat: graph-edges from linked_entities + BrainBench sealed corpus | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1467 | feat(reranker): rewire cross-encoder judge to CF US open-weight model | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1466 | chore(prod): activate cross-encoder reranker on production (FIN_CROSS_ENCODER_RERANK_ENABLED=true) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1464 | docs: record QC production sync proof | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1463 | QC social-photo inspection E2E proof | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1462 | feat(retrieval): real LLM-judge cross-encoder scorer + wire it so rerank reorders (gbrain-parity P1-L2) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1461 | feat(prod-brain): entity_doc_projection seeder (operational, no-execute) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1460 | fix(p0): reembed worker — pre-filter non-embeddable input + per-item batch isolation | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1459 | feat: Wave-2 batch — weather mirror + tenant→central feedback + skill breadth batch 2 | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1458 | feat(eval): close 3 of 4 gated-suite canonicals with real-event invariants | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1457 | fix(dream-cycle): per-phase savepoint isolation so one phase throw can’t zero the pass (prod revival) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1456 | fix(cron): consolidated cron-fixes batch — commodity-mirror Worker-conn + eval-runner emit-resilience + Worker observability | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1455 | feat(p2+p5): Wave-1b clean batch — think-op wrapper + corpus-env hygiene + skill-breadth + mode-bundles | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1454 | feat(p2+p4+p5): Wave-1 depth — python eval bridge + routing resolver + schema-pack authority + freshness SLA | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1453 | activate(prod): enable commodity-profile mirror (FIN_COMMODITY_MIRROR_ENABLED) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1452 | feat(p4): central→tenant commodity-profile mirror (the correct two-plane freshness fix) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1451 | activate(prod): validation activation — max-pool + dream-enrich + LSD-apply + gated-suite-eval-runner | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1450 | feat(p5): autonomous enrichment — dream-cycle enrich phase + think/synthesis op | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1449 | feat(p4): bounded eval-gated Central-Hermes execute authority (owner-granted, inert) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1448 | Release QC social photo inspection and Nathel auth proof | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1447 | feat(p5): derived-enrichment substrate + immutability fence | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1446 | feat(p2+p3): eval-gate robustness (judge fingerprint + anti-gaming) + operator verdict reader | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1445 | feat(p2): enforce-readiness — eval-gate suite-id resolution + gated-suite eval-result emitters | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1444 | fix(p0): token-safe embedding chunker — never exceed the 8192-token model limit | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1443 | feat(p3): close the loops — LSD-supersession applier + tenant-intel producer/consumer + operator authorization wiring | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1442 | feat(p3): close-the-loop prep — deriveExecutionAuthorization + resolveConfig live + mission-packet pre-gate + knobs_hash cache-key | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1441 | feat(p1+p2): max-pool embedding + cross-encoder reranker + query-anchor graph seed + canonical eval suite-id registry | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1440 | feat(p0): prod-brain data-plane operational scripts (apply-065/066 runbook, backlink backfill, reembed bulk-drain, coverage verifier, graph probe) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1439 | feat(mcp): register find_trajectory + wire it as the cheap-signal gate before the LSD judge | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1438 | feat(activation): graph-channel + schema-pack (both envs) + eval-gate shadow (prod) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1437 | fix(retrieval): fail-open guard on graph RRF channel (42P01 → vector+lexical) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1436 | activate(prod): R1 live-harvest daily cron (un-starves the ingestion/graph cluster) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1435 | P2 retrieval: deterministic query expansion (synonym/alias) + activate | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1434 | P5 LSD: wire contradiction-judge into the dream-cycle + activate (+ R1 cron guard) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1432 | feat(skill-invoker): real execution recipe — propose→act genuinely closes | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1431 | activate(prod) Tier 2: structuring/serving/eval engines (11 audited-safe flags) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1430 | activate(prod) Tier 1: read-path retrieval/routing channels (owner-approved) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1429 | Activate eval-gate SHADOW on staging (close the EVAL/HEAL seam) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1428 | P3: composition_skill producer — complete the mentor loop end-to-end | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1427 | P3: mentor Evidence-Pack producer wiring (shadow, flag-gated) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1426 | P6: tenant→central intel proposal emitter (proposal-only, flag-gated) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1425 | test(qc): expand frontier state lab and repair Nathel role proof | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1424 | P3+P5: mentor response-composition + LSD trio (BudgetTracker, find_trajectory, contradiction-judge) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1423 | activate(staging): enable landed engine flags (central Hermes, skill-invoker, self-wiring, dream-cycle, source-registry) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1422 | Next-wave batch: tenant skill-invoker + mentor Evidence Pack + self-wiring graph | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1421 | feat(qc): refine social photo inspection composer | robertolouisgarcia-maker |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#34 | fix(contracts): sync GDF sha256 to governed hash | robertolouisgarcia-maker |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#33 | fix(contracts): sync GDF schema to governed fin-agentic copy (cross-repo drift) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1420 | P1-P4 batch: Central Hermes + pure-mirror + vendor-label gate + nDCG eval + retrieval modularize | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1419 | fix(p0): re-seed eval-gate personas in staging deploy — unblock the activation backlog | robertolouisgarcia-maker |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#32 | feat(c1): wire the raw-lake → central-corpus structuring loop (the root unlock) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1418 | docs(program): FIN Agentic ERP Execution Program (2026-06-22) — the framed plan | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1417 | Fix QC simulator auth proof harness | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1416 | fix(cron): un-strand heartbeat + QC wiki-writer crons in production (flag-gated) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1415 | fix(sse): flush initial heartbeat before cold-start tip query — unblocks staging deploy gate | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1414 | enable(retrieval): named-thing channel ON for staging (measured vendor-resolution 4/4) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1413 | Retrieval: named-thing channel — vendor-resolution 0/4→4/4, overall P@5 0.27→0.45 (MEASURED) | robertolouisgarcia-maker |
fin-platform/fin-agentic | fin-platform/fin-agentic#1408 | Fix Nathel QC Xcode simulator auth | robertolouisgarcia-maker |
fin-platform/knowledge-forge | fin-platform/knowledge-forge#31 | feat(kb-freshness): sources.yaml load-bearing + Exa crawler (T1 slice 1) | robertolouisgarcia-maker |
By the numbers
- PRs opened (work started): 217
- PRs merged (work completed): 213
- Issues closed (work completed): 0
- Commits (supporting signal): 213
- Repositories touched: 2
| Repo | Opened | Merged | Closed | Commits |
|---|---|---|---|---|
fin-platform/fin-agentic | 214 | 209 | 0 | 209 |
fin-platform/knowledge-forge | 3 | 4 | 0 | 4 |
Previous period: github_activity_digest_weekly_2026_W25.
Links
- fin-platform/fin-agentic#1626 (opened)
- fin-platform/fin-agentic#1625 (opened)
- fin-platform/fin-agentic#1624 (opened)
- fin-platform/fin-agentic#1623 (opened)
- fin-platform/fin-agentic#1622 (opened)
- fin-platform/fin-agentic#1621 (opened)
- fin-platform/fin-agentic#1620 (opened)
- fin-platform/fin-agentic#1619 (opened)
- fin-platform/fin-agentic#1618 (opened)
- fin-platform/fin-agentic#1617 (opened)
- fin-platform/fin-agentic#1616 (opened)
- fin-platform/fin-agentic#1615 (opened)
- fin-platform/fin-agentic#1614 (opened)
- fin-platform/fin-agentic#1613 (opened)
- fin-platform/fin-agentic#1612 (opened)
- fin-platform/fin-agentic#1611 (opened)
- fin-platform/fin-agentic#1610 (opened)
- fin-platform/fin-agentic#1609 (opened)
- fin-platform/fin-agentic#1608 (opened)
- fin-platform/fin-agentic#1607 (opened)
- fin-platform/fin-agentic#1606 (opened)
- fin-platform/fin-agentic#1605 (opened)
- fin-platform/fin-agentic#1604 (opened)
- fin-platform/fin-agentic#1603 (opened)
- fin-platform/fin-agentic#1602 (opened)
- fin-platform/fin-agentic#1601 (opened)
- fin-platform/fin-agentic#1600 (opened)
- fin-platform/fin-agentic#1599 (opened)
- fin-platform/fin-agentic#1598 (opened)
- fin-platform/fin-agentic#1597 (opened)
- fin-platform/fin-agentic#1596 (opened)
- fin-platform/fin-agentic#1595 (opened)
- fin-platform/fin-agentic#1594 (opened)
- fin-platform/fin-agentic#1593 (opened)
- fin-platform/fin-agentic#1592 (opened)
- fin-platform/fin-agentic#1591 (opened)
- fin-platform/fin-agentic#1590 (opened)
- fin-platform/fin-agentic#1589 (opened)
- fin-platform/fin-agentic#1588 (opened)
- fin-platform/fin-agentic#1587 (opened)
- fin-platform/fin-agentic#1586 (opened)
- fin-platform/fin-agentic#1585 (opened)
- fin-platform/fin-agentic#1584 (opened)
- fin-platform/fin-agentic#1583 (opened)
- fin-platform/fin-agentic#1582 (opened)
- fin-platform/fin-agentic#1581 (opened)
- fin-platform/fin-agentic#1580 (opened)
- fin-platform/fin-agentic#1579 (opened)
- fin-platform/fin-agentic#1578 (opened)
- fin-platform/fin-agentic#1577 (opened)
- fin-platform/fin-agentic#1576 (opened)
- fin-platform/fin-agentic#1575 (opened)
- fin-platform/fin-agentic#1574 (opened)
- fin-platform/fin-agentic#1573 (opened)
- fin-platform/fin-agentic#1572 (opened)
- fin-platform/fin-agentic#1571 (opened)
- fin-platform/fin-agentic#1570 (opened)
- fin-platform/fin-agentic#1569 (opened)
- fin-platform/fin-agentic#1568 (opened)
- fin-platform/fin-agentic#1567 (opened)
- fin-platform/fin-agentic#1566 (opened)
- fin-platform/fin-agentic#1565 (opened)
- fin-platform/fin-agentic#1564 (opened)
- fin-platform/fin-agentic#1563 (opened)
- fin-platform/fin-agentic#1562 (opened)
- fin-platform/fin-agentic#1561 (opened)
- fin-platform/fin-agentic#1560 (opened)
- fin-platform/fin-agentic#1559 (opened)
- fin-platform/fin-agentic#1558 (opened)
- fin-platform/fin-agentic#1557 (opened)
- fin-platform/fin-agentic#1556 (opened)
- fin-platform/fin-agentic#1555 (opened)
- fin-platform/fin-agentic#1554 (opened)
- fin-platform/fin-agentic#1553 (opened)
- fin-platform/fin-agentic#1552 (opened)
- fin-platform/fin-agentic#1551 (opened)
- fin-platform/fin-agentic#1550 (opened)
- fin-platform/fin-agentic#1549 (opened)
- fin-platform/fin-agentic#1548 (opened)
- fin-platform/fin-agentic#1547 (opened)
- fin-platform/fin-agentic#1546 (opened)
- fin-platform/fin-agentic#1545 (opened)
- fin-platform/fin-agentic#1544 (opened)
- fin-platform/fin-agentic#1543 (opened)
- fin-platform/fin-agentic#1542 (opened)
- fin-platform/fin-agentic#1541 (opened)
- fin-platform/fin-agentic#1540 (opened)
- fin-platform/fin-agentic#1539 (opened)
- fin-platform/fin-agentic#1538 (opened)
- fin-platform/fin-agentic#1537 (opened)
- fin-platform/fin-agentic#1536 (opened)
- fin-platform/fin-agentic#1535 (opened)
- fin-platform/fin-agentic#1534 (opened)
- fin-platform/fin-agentic#1533 (opened)
- fin-platform/fin-agentic#1532 (opened)
- fin-platform/fin-agentic#1531 (opened)
- fin-platform/fin-agentic#1530 (opened)
- fin-platform/fin-agentic#1529 (opened)
- fin-platform/fin-agentic#1528 (opened)
- fin-platform/fin-agentic#1527 (opened)
- fin-platform/fin-agentic#1526 (opened)
- fin-platform/fin-agentic#1525 (opened)
- fin-platform/fin-agentic#1524 (opened)
- fin-platform/fin-agentic#1523 (opened)
- fin-platform/fin-agentic#1522 (opened)
- fin-platform/fin-agentic#1521 (opened)
- fin-platform/fin-agentic#1520 (opened)
- fin-platform/fin-agentic#1519 (opened)
- fin-platform/fin-agentic#1518 (opened)
- fin-platform/fin-agentic#1517 (opened)
- fin-platform/fin-agentic#1516 (opened)
- fin-platform/fin-agentic#1515 (opened)
- fin-platform/fin-agentic#1514 (opened)
- fin-platform/fin-agentic#1513 (opened)
- fin-platform/fin-agentic#1512 (opened)
- fin-platform/fin-agentic#1511 (opened)
- fin-platform/fin-agentic#1510 (opened)
- fin-platform/fin-agentic#1509 (opened)
- fin-platform/fin-agentic#1508 (opened)
- fin-platform/fin-agentic#1507 (opened)
- fin-platform/fin-agentic#1506 (opened)
- fin-platform/fin-agentic#1505 (opened)
- fin-platform/fin-agentic#1504 (opened)
- fin-platform/fin-agentic#1503 (opened)
- fin-platform/fin-agentic#1502 (opened)
- fin-platform/fin-agentic#1501 (opened)
- fin-platform/fin-agentic#1500 (opened)
- fin-platform/fin-agentic#1499 (opened)
- fin-platform/fin-agentic#1498 (opened)
- fin-platform/fin-agentic#1497 (opened)
- fin-platform/fin-agentic#1496 (opened)
- fin-platform/fin-agentic#1495 (opened)
- fin-platform/fin-agentic#1494 (opened)
- fin-platform/fin-agentic#1493 (opened)
- fin-platform/fin-agentic#1492 (opened)
- fin-platform/fin-agentic#1491 (opened)
- fin-platform/fin-agentic#1490 (opened)
- fin-platform/fin-agentic#1489 (opened)
- fin-platform/fin-agentic#1488 (opened)
- fin-platform/fin-agentic#1487 (opened)
- fin-platform/fin-agentic#1486 (opened)
- fin-platform/fin-agentic#1485 (opened)
- fin-platform/fin-agentic#1484 (opened)
- fin-platform/fin-agentic#1483 (opened)
- fin-platform/fin-agentic#1482 (opened)
- fin-platform/fin-agentic#1481 (opened)
- fin-platform/fin-agentic#1480 (opened)
- fin-platform/fin-agentic#1479 (opened)
- fin-platform/fin-agentic#1478 (opened)
- fin-platform/fin-agentic#1477 (opened)
- fin-platform/fin-agentic#1476 (opened)
- fin-platform/fin-agentic#1475 (opened)
- fin-platform/fin-agentic#1474 (opened)
- fin-platform/fin-agentic#1473 (opened)
- fin-platform/fin-agentic#1472 (opened)
- fin-platform/fin-agentic#1471 (opened)
- fin-platform/fin-agentic#1470 (opened)
- fin-platform/fin-agentic#1469 (opened)
- fin-platform/fin-agentic#1468 (opened)
- fin-platform/fin-agentic#1467 (opened)
- fin-platform/fin-agentic#1466 (opened)
- fin-platform/fin-agentic#1465 (opened)
- fin-platform/fin-agentic#1464 (opened)
- fin-platform/fin-agentic#1463 (opened)
- fin-platform/fin-agentic#1462 (opened)
- fin-platform/fin-agentic#1461 (opened)
- fin-platform/fin-agentic#1460 (opened)
- fin-platform/fin-agentic#1459 (opened)
- fin-platform/fin-agentic#1458 (opened)
- fin-platform/fin-agentic#1457 (opened)
- fin-platform/fin-agentic#1456 (opened)
- fin-platform/fin-agentic#1455 (opened)
- fin-platform/fin-agentic#1454 (opened)
- fin-platform/fin-agentic#1453 (opened)
- fin-platform/fin-agentic#1452 (opened)
- fin-platform/fin-agentic#1451 (opened)
- fin-platform/fin-agentic#1450 (opened)
- fin-platform/fin-agentic#1449 (opened)
- fin-platform/fin-agentic#1448 (opened)
- fin-platform/fin-agentic#1447 (opened)
- fin-platform/fin-agentic#1446 (opened)
- fin-platform/fin-agentic#1445 (opened)
- fin-platform/fin-agentic#1444 (opened)
- fin-platform/fin-agentic#1443 (opened)
- fin-platform/fin-agentic#1442 (opened)
- fin-platform/fin-agentic#1441 (opened)
- fin-platform/fin-agentic#1440 (opened)
- fin-platform/fin-agentic#1439 (opened)
- fin-platform/fin-agentic#1438 (opened)
- fin-platform/fin-agentic#1437 (opened)
- fin-platform/fin-agentic#1436 (opened)
- fin-platform/fin-agentic#1435 (opened)
- fin-platform/fin-agentic#1434 (opened)
- fin-platform/fin-agentic#1433 (opened)
- fin-platform/fin-agentic#1432 (opened)
- fin-platform/fin-agentic#1431 (opened)
- fin-platform/fin-agentic#1430 (opened)
- fin-platform/fin-agentic#1429 (opened)
- fin-platform/fin-agentic#1428 (opened)
- fin-platform/fin-agentic#1427 (opened)
- fin-platform/fin-agentic#1426 (opened)
- fin-platform/fin-agentic#1425 (opened)
- fin-platform/fin-agentic#1424 (opened)
- fin-platform/fin-agentic#1423 (opened)
- fin-platform/fin-agentic#1422 (opened)
- fin-platform/fin-agentic#1421 (opened)
- fin-platform/knowledge-forge#34 (opened)
- fin-platform/knowledge-forge#33 (opened)
- fin-platform/fin-agentic#1420 (opened)
- fin-platform/fin-agentic#1419 (opened)
- fin-platform/knowledge-forge#32 (opened)
- fin-platform/fin-agentic#1418 (opened)
- fin-platform/fin-agentic#1417 (opened)
- fin-platform/fin-agentic#1416 (opened)
- fin-platform/fin-agentic#1415 (opened)
- fin-platform/fin-agentic#1414 (opened)
- fin-platform/fin-agentic#1413 (opened)
- fin-platform/fin-agentic#1408 (merged)
- fin-platform/knowledge-forge#31 (merged)
Provenance
Auto-generated by the scripts/maint/generate-gh-digest@1.0.0 projector from gh query output.
- Owner scope:
fin-platform(all active FIN repos) - Window:
2026-06-22T00:00:00Z→2026-06-29T00:00:00Z - Sources:
github:prs.opened,github:prs.merged,github:issues.closed,github:commits - Lede + Highlights: LLM editor (
LLMProvider) - Regenerate:
pnpm digest:gh --type weekly --date 2026-06-22