memarium-recalllisted

**TRIGGER IMMEDIATELY when the user's question contains retrospective phrasing in ANY language:** 之前是怎么解的 / 之前怎么解的 / 上次怎么处理的 / 上次怎么解决的 / 以前遇到过吗 / 以前我们怎么做 / 之前我们试过吗 / how did we fix X before / how was Y solved / why does Z work this way / what did we try last time / did we already try W. **Also trigger on:** design/architecture/prior-art questions — what pattern should I use for X / how have we approached Y before / what did we decide about Z / is there prior art for W. **Anti-pattern to break before reflex kicks in:** the question often contains clear keywords (file name, crash type, API) and your shortcut instinct is `git log --grep="<keyword>"` — DO NOT. Commit messages strip the conversation context where the user explained what didn't work and why; memarium's typed memory preserves decisions, dead ends, trade-offs, and unmerged-attempt context that git history can't surface. Run memarium-recall (one CLI call, ~small) FIRST; if nothing matches, *then* fall back to git. Two-stage recall over typed memory —
june9593/memarium-plugin · ★ 1 · Code & Development · score 72

Install: claude install-skill june9593/memarium-plugin

# /memarium-recall — read your own notes before doing the work You (in-session Claude) just landed in a project repo. The user has been working in this repo (and others) for weeks/months, and the memarium plugin has captured every Claude Code + Copilot + supported Codex session into `~/.memarium/session-repo/`. The `/memarium` digest distilled those sessions into **typed memory** — `episodic` (the arc / dead-ends / decisions of a past work thread), `semantic` (durable project facts), `procedural` (how-to playbooks + gotchas), `core` (never-forget rules) — one AI-native, scored knowledge layer. (There is no human "book" anymore.) **Your job here**: before you explore code, run recall with your task keywords, then `Read` the top 1–5 hits. Past-you may have already debugged exactly this. ## Two-stage recall (low context by design) | Stage | Payload | Question it answers | |---|---|---| | 1 (this command) | ranked hits — title + 1-line summary + `whyRecalled` per entry, bodies NOT loaded | "Which past episodes/facts/procedures bear on my task?" | | 2 (`Read` tool) | full body, ~1–3 KB per entry | "What did past-me actually do / decide / hit here?" | The CLI does the ranking (keyword overlap over title/summary/entities + file/commit overlap + recency + importance), so you don't triage a topic tree — you get a relevance-sorted list directly. ## Step 0 — locate the plugin binary ```bash VBP="${CLAUDE_PLUGIN_ROOT:+$CLAUDE_PLUGIN_ROOT/bin/memarium-plugin.js}" [ -x "$VBP" ] || V