skills/use-cli-recall/SKILL.md
Search past Claude Code/Codex sessions via recall CLI.
npx skillsauth add thkt/dotclaude use-cli-recallInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
When one applies, call without deliberation.
| Trigger | Signal | | -------------------- | ------------------------------------------ | | Temporal reference | 「前に」「あの時」 past events / decisions | | Structural echo | Current problem mirrors a past situation | | Repetition | 「また同じ」 recurring mistake | | Vague back-reference | 「あの件」 past work without specifics | | Module first contact | First edit to a file/module this session |
| Purpose | Command |
| ----------------- | ----------------------------------------------------------- |
| Search | recall search "query". Shorthand: recall "query" |
| Last N days | recall search "query" --days N |
| Project filter | recall search "query" --project <path> |
| Source filter | recall search "query" --source claude or --source codex |
| Limit results | recall search "query" --limit N. Default 10, max 100 |
| Show session | recall show <session-id> |
| Status | recall status |
| Incremental index | recall index |
| Full rebuild | recall index --force |
Write bilingual queries upfront (e.g. recall "認証 auth"). FTS5 trigram tokenization cannot match JA terms of 2 chars or fewer; 認証 and 依存 hit 0. Embeddings do not bridge EN⇄JA (thkt/recall#51). Including both languages covers each search path.
recall does not expand queries (caller-is-LLM, thkt/recall#25). Hybrid search returns nearest neighbors, so a poor query yields low-relevance results rather than 0 hits. When results are empty or low-relevance, rewrite the query yourself and retry once: synonyms, EN⇄JA variants, related concept terms.
use-cli-recall answers past decisions; ugrep / bfs answer the current code state. Run both on these triggers.
| Trigger | recall query | code search | | -------------------- | ----------------------------- | ---------------------------- | | Module first contact | module name, design rationale | module name, key identifiers | | Structural echo | past similar problem | current similar code |
tools
Delegate implementation to codex (coder) via the herdr-agentchat plugin and drive a two-pane conversation to completion.
development
Extract recurring patterns from past closed PRs/issues and the research findings in workspace/research/, verify them against the latest code, and propose them to docs/wiki/ via PR.
development
Create Decision Records (DR) in MADR v4 format with auto-numbering.
development
Detect flaky tests by shaking them — repeated runs under varied order, parallelism, and seed — plus a static smell scan that flags latent flakiness in tests that currently pass. Classify each target as confirmed-flaky, latent-flaky, or stable and fix the root cause without weakening the test. Do NOT use to fix a confirmed single bug (use /fix) or for static-only code review (use /audit).