graphrag-stocktake

name	graphrag-stocktake
description	vault の Investigation ライフサイクル棚卸し(定期クリーニング)。state無しレガシーや、決着済みなのに active のままの focus を、機械検出(stocktake verb)+ 裏取り裁定で state:closed に整える。「棚卸しして」「vault を掃除/クリーニングして」「Investigation を整理して」「active が溜まってる」で発火。resume の stocktake_hint が出た時にも。削除はしない(閉じるだけ)。スラッシュ: /graphrag-knowledge:graphrag-stocktake

Investigation stocktake (lifecycle periodic cleaning)

A periodic sweep that tidies the state of Investigations accumulated in the vault, via machine detection + corroborated adjudication. For the read/write foundation and CLI details, follow the parent skill graphrag-knowledge and $REF/ (= ${CLAUDE_PLUGIN_ROOT}/references). $CLI = node --experimental-strip-types ${CLAUDE_PLUGIN_ROOT}/graphrag/cli.ts.

Scope (what this skill does and does not do)

• The primary channel for cleanup is the always-on triggers (pre-commit nudge, checkpoint, the closed judgment at write-back when implementation reaches a milestone). This skill is the later-stage periodic sweep that picks up their leftovers/misses, not a replacement for the always-on triggers.
• Detection is delegated to the deterministic verb (stocktake); this skill handles adjudication only. Do not make the LLM read the whole vault (per §Anti-patterns in the parent skill).
• Fire on an explicit instruction like "stocktake", or when you see the stocktake_hint returned by brief --mode resume.

Procedure

1. Run $CLI stocktake [--days <N=14>] [--vault <dir>] to get the suspect list (read-only, deterministic). signals: stateless | stale-active | no-generated-at | progress-claim. If suspects are zero, report "healthy" and stop.
2. Adjudicate each suspect with corroboration. Adjudication rules (the core of this skill):
   • Corroboration against code, tests, git evidence, and chat evidence beats the summary's self-report (unimplemented / in-progress / incomplete). If tests or an implementation actually exist / it shipped, the investigation is settled = closed even if the summary says "unimplemented".
   • Remaining code work is owned by the Decision/Risk side — an Investigation closes on "the investigation / focus being settled". Even if the code is unfinished, if the investigation itself (what was tried and what was learned) is done, closed is fine.
   • A genuinely ongoing focus stays active (if a stateless suspect is really still a moving focus, explicitly assign state:"active" — do not leave a dangling item found during stocktake unattended).
   • If you cannot corroborate and are unsure, do not close it — report it (do not close on your own). A state change with no basis leads to irreversible information loss.
3. Bundle the settled ones into a single commit-mutation (a plan that changes only state via op:update) and apply in one batch. For just a single item, a minimal plan on the spot suffices.
4. Report (per §Reporting format in the parent skill): in natural language, briefly — the count closed and a summary of the grounds / the focuses left active / the deferrals and their reasons. Do not dump IDs / raw JSON.

Off-limits (never do)

• Do NOT delete. Lineage preservation is the way — closed is discounted 0.6× in ranking, so it naturally sinks in search results (it is not excluded).
• Do NOT step into whether knowledge nodes (Decision/Risk/OK, etc.) are "still true" — that is the domain of drift-reconciliation ($REF/drift-reconciliation.md). This skill handles only the state of Investigations.