memhub

Read-only health check of the memhub project in this repo — schema version, render freshness, write-log activity

Bootstrap memhub in this repo — initialize the SQLite store, seed starter state and architecture narratives, render PROJECT.md and PROJECT_LEDGER.md

Initialize memhub for a repo in OpenCode; use when the user asks to bootstrap project memory.

Summarize this memhub session, route updates into the database, then re-render PROJECT.md and PROJECT_LEDGER.md

Rebuild + install memhub and bring every memhub instance on this machine (each known repo DB + the machine-global store) to head schema, resync installed agent skill wrappers, with a one-time fix for the ~/.local/bin PATH shadow. Run from the memhub source repo.

Wrap up an OpenCode memhub session; run the binary-rendered policy, draft updates, get per-item approval, write, and render.

Run memhub's read-only CLAUDE.md/AGENTS.md drift-and-bloat audit (`memhub audit md`) and interpret the findings — size, generated-file drift, managed-block version, keystone phrases, opt-in user-global size — recommending concrete fixes without rewriting anything itself. Trigger on: "audit CLAUDE.md", "check for CLAUDE.md drift", "is CLAUDE.md bloated", "run the md audit", "/audit-md".

Run the memhub Recall@K eval harness against tests/retrieval_golden.json and report the baseline. Read-only; never mutates the DB or writes_log.

Run and interpret memhub's read-only CLAUDE.md/AGENTS.md drift-and-bloat audit in OpenCode; use when the user asks whether CLAUDE.md is bloated, drifted, or out of date. Trigger on: "audit CLAUDE.md", "check for CLAUDE.md drift", "is CLAUDE.md bloated", "run the md audit", "/audit-md".

Run the memhub retrieval eval harness from OpenCode; use to check Recall@K quality.

Look up relevant facts, decisions, and tasks for the current conversation via memhub recall (SQL+RAG hybrid). Prefer this over reading PROJECT_LEDGER.md mid-session. Trigger on: "what did we decide about X", "is there a fact/decision/task about Y", "recall X", "what do we know about Z", "look this up in memhub".

Find where code lives in this repo by intent — SQL+RAG hybrid search over a sibling code index that returns ranked file:line breadcrumbs with clipped snippets. Read-only; never returns full code. Trigger on: "where is X", "where does X live", "find the code that does Y", "which file handles Z", "locate X", "where do I change Y".

Locate code in this repo by intent in OpenCode; use when you need to find where a symbol, function, or behavior lives before reading or editing it. Trigger on: "where is X", "where does X live", "find the code that does Y", "which file handles Z", "locate X", "where do I change Y".

Check memhub project health in OpenCode; use when the user asks whether memhub is initialized or healthy.

Ingest an external markdown reference doc into memhub so it is RAG-searchable in chunks. The first ingest in a repo auto-enables doc chunks in default recall (decision 90). Trigger on: "ingest this doc", "add this spec to memhub", "make this doc searchable", "remember this reference doc", "/doc".

Recall memhub project memory in OpenCode; use when project facts, decisions, tasks, or docs are needed mid-session. Trigger on: "what did we decide about X", "is there a fact/decision/task about Y", "recall X", "what do we know about Z", "look this up in memhub".

Show the token-accounting dashboard panel for the current memhub project — 7-day and 30-day recall/session token totals, context offset vs full-ledger baseline, and a recent-sessions table. Trigger on: "token usage", "context cost", "how many tokens", "recall savings", "show memhub metrics", "token accounting".

Ingest markdown reference docs into memhub from OpenCode; use for design specs, API contracts, and guides. Trigger on: "ingest this doc", "add this spec to memhub", "make this doc searchable", "remember this reference doc", "/doc".

Show memhub token metrics in OpenCode; use when the user asks about recall/session token accounting. Trigger on: "token usage", "context cost", "how many tokens", "recall savings", "show memhub metrics", "token accounting".

Pull this repo's memhub DB from the Google Drive synced folder, compare it to local, and (with user approval) adopt the newer state so this machine has memory from sessions on other machines.

Pull this repo's memhub DB from the Google Drive synced folder at session start and adopt newer cross-machine state with user approval.

Manage memhub machine-global memory in OpenCode; use for cross-repo facts, decisions, and docs.

Rebuild memhub embeddings in OpenCode; use after stale-embedding warnings or embedding model changes.

Upgrade memhub and resync agent wrappers from OpenCode; use after code changes or stale installed skills.

Launch the memhub dashboard from OpenCode; use when the user asks for the local visual memory dashboard.

Manage memhub machine-global memory — opt this repo in/out, write or promote facts/decisions/docs that should be visible to every repo on this machine, and understand scope provenance in recall.

Launch the memhub viz dashboard for the current repo in a browser tab.

Rebuild memhub embeddings for the active model. Use after a model upgrade or when memhub.recall surfaces a stale_embeddings warning. Always ask before running.