memd¶
memd is a local memory CLI that gives coding agents and AI scientists a
single shared, persistent memory — raw searchable content, structured task
history, and canonical collaboration artifacts — indexed by a hybrid dense +
sparse retrieval stack and gated by an explicit trust boundary.
Agents retrieve bounded context with memd agent-context or memd search,
read the generated context file, and record durable progress with memd add.
For low-latency local use, memd keeps the store and indexes hot through a
private CLI-managed warm worker driven by ordinary CLI commands.
What memd does¶
| Surface | Purpose | Primary CLI commands |
|---|---|---|
| Raw memory | Store and search chunks: code, docs, notes, traces, decisions | memd add, memd search, memd get, memd stats |
| Agent context | Bounded pre-work context + JSON audit logs | memd agent-context --output .memd/context.md |
| Warm CLI | Keep store/index state hot for repeated local calls | memd warm start, memd warm status |
| Batch CLI | Many structured operations in one loaded process | memd batch --jsonl requests.jsonl |
| Export/import | Move local memory to portable formats | memd export-omf, memd import-omf |
| Operations | Structured memory / task / artifact / context / code / debug ops | memd call task.start --json '{…}' |
| Guardrails | Pin tenant/project scope and verify CLI-first agent wiring | memd init, memd doctor |
Headline number¶
Cross-system LoCoMo retrieval (10 conversations, 5,882 turns, 1,536 queries):
| System | MRR@10 | Hit@10 | Avg search |
|---|---|---|---|
memd v0.50.0 |
0.420 | 0.621 | 26.7 ms |
superlocalmemory v3.4.46 (lexical) |
0.369 | 0.599 | 804.5 ms |
mem0 v2.0.2 (LLM-extracted) |
0.354 | 0.591 | 40.9 ms |
memd wins on every quality metric and search latency. See
Benchmarking for per-category breakdown, reproducibility
notes, and the internal-corpus + Bright-Pro adapter results.
Start here¶
- Quick start — install, store first memory, retrieve, build agent context.
- Operational contract — what agents should write, avoid, verify, and clean up.
- Architecture — hybrid retrieval, storage, trust boundary diagram.
- CLI reference — every command and operation.
- Agent skill — install
memdinto Claude Code / Codex with one command.