[v6-p4p5] Recursive planner (refuses unsafe, proposes one, stops) + soak operations (889 tests green)#50
Merged
Merged
Conversation
…priority + cycle ledger (red) https://claude.ai/code/session_01AgdV9SKZZP6JbyTBo2gAWZ
…ts PlanCard, stops at the proposal (pnpm loop:plan) https://claude.ai/code/session_01AgdV9SKZZP6JbyTBo2gAWZ
…s CLI + SOAK_OPERATIONS runbook https://claude.ai/code/session_01AgdV9SKZZP6JbyTBo2gAWZ
…P3 operator real-proof gap (top safety_evidence) https://claude.ai/code/session_01AgdV9SKZZP6JbyTBo2gAWZ
…5 code complete); session log s_v6_0003 https://claude.ai/code/session_01AgdV9SKZZP6JbyTBo2gAWZ
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
V6-P4 (recursive planner) + V6-P5 (soak operations)
Recursive planner (
packages/daemon/src/recursive-planner.ts, 20 tests): pureplanNextCyclewith 8 refusal paths (dirty tree / red tests / blocked budget / ambiguous SoT / open holds / unmerged prev PR / unparseable §0 / empty registry — each with human recovery); strict single-pick prioritysafety_evidence > user_ux > automation > fleet > polish; cycle ledger toevidence/loop-cycles/AND event-sourced (planner.cycle_planned, rebuildable — GR#5).pnpm loop:planshell gathers real inputs (actually runs the test suite, scans SoT claimants, checks the real budget db) and stops at the proposal — never implements, never pushes (GR#10).First real run committed as evidence: a smoke run honestly REFUSED on a dirty tree; the official clean run proposed cycle 1 =
v6-p3-real-proof-closeout(safety_evidence) — exactly the operator-gated top gap.evidence/loop-cycles/cycle-1.json.Soak ops (
docs/operations/SOAK_OPERATIONS.md+pnpm soak:status, 5 tests): one-week command, launchd plist, evidence contract, resume/failure recovery, ntfy wiring,soak-pending.jsonstatus artifact (overdue detection, sticky terminal states).WORKBOOK_v6 §0 → V6-P6; V6-P3 stays honestly operator-gated. Gates: typecheck ✅ lint ✅ test 889 passed / 0 failed / 6 skipped (+25).
https://claude.ai/code/session_01AgdV9SKZZP6JbyTBo2gAWZ
Generated by Claude Code