CLI Reference
memax recall
Semantic search across your memories. Returns the most relevant pieces of knowledge for a given query.
Semantic search for relevant memories. This is the primary retrieval command — used by agents and humans to find context.
Usage
memax recall <query> [options]Options
| Flag | Description | Default |
|---|---|---|
--limit <n> | Maximum results | 5 |
--hub <id> | Scope to a specific hub | All accessible hubs |
--hint <text> | Additional context for better ranking | — |
--threshold <n> | Minimum relevance score (0–1) | 0.3 |
--format json | Output as JSON | Human-readable |
--quiet | Output content only, no metadata | — |
Examples
Basic recall
memax recall "how does authentication work?"─── Auth System Design (core · private) ─────────────────────
The auth system uses JWT with short-lived access tokens (1h)
and refresh tokens (30d). OAuth2 via GitHub and Google...
Relevance: 0.94 · Pushed 3 days ago
─── Login Flow (reference · team) ───────────────────────────
Users authenticate via /api/auth/login. The handler validates
credentials, generates a token pair, and sets an HTTP-only
refresh cookie...
Relevance: 0.87 · Pushed 1 week agoWith context hints
memax recall "token refresh" --hint "debugging a 401 error in production"JSON output for scripting
memax recall "deploy process" --format json | jq '.[0].content'Quiet mode for piping
# Pipe directly to an agent
memax recall "coding conventions" --quiet | pbcopyHow it works
- Your query is embedded into a vector
- pgvector finds the most similar chunks
- Cohere Rerank scores each candidate
- Results are filtered by boundary and deduplicated
- Top results are returned with relevance scores
The entire pipeline runs in under 500ms (p95).