The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Codemem listing page.
Your coding agents keep rebuilding what you already have. codemem remembers what they built, across every project and every machine, and scores each piece on evidence of whether it worked.
It is an MCP server for Claude Code and other MCP clients. It knows what projects exist, where they live, what reusable pieces they contain, what each session did and decided, every commit pushed to the git host, and how to do recurring things. Everything is searchable from any machine, from an agent, or from the web UI. It is written to during work, not after.
One Python process, SQLite, no external services required.
Inventory scripts snapshot a directory and you read the report once. Registries list what people published. codemem does neither.
verify records
"I ran it today and it works" and restarts the freshness clock. Scores are recomputed on
every sync, so they decay when you stop touching something.authoritative, usable, experimental, antiquated,
sunset, broken, junk, each with a reason. Search ranks by it, and you can exclude the
ratings you should not build on.| Endpoint | What |
|---|---|
http://<host>:8055/mcp | MCP (streamable HTTP) for Claude Code and other MCP clients |
http://<host>:8055/ | Web UI: search, projects, assets, notes, activity, docs. Dark mode. |
http://<host>:8055/api/… | JSON API used by the UI (see docs/architecture.md) |
http://<host>:8055/health | liveness |
Normally runs on the git host, the machine holding the bare repos, so it can read them directly.
Security: codemem has no authentication. It is built for a trusted network, and anyone who can reach the port can read and write everything. Do not expose it to the internet without an authenticating proxy in front of it. See docs/architecture.md.
Python 3.11 or newer, and git. Optional: a git host with bare repos for the commit feed, Gitea for descriptions and web links, and Ollama for semantic search, drafted descriptions and the review pass. Everything works without the optional pieces; search falls back to BM25 alone.
A developer with many project folders across several machines has no way for a new Claude Code session to know that a retry wrapper, a Gradio helper, a GPU monitor or a data pipeline already exists somewhere. So they get rebuilt. Snapshot inventory scripts (run, read, forget) do not fix that. codemem is the live, writable layer: it is written to during work, not after.
The content is whatever is there. codemem does not filter or judge. Each project carries an
audience: unrestricted (the default), professional, or employer, so a search or a view can
exclude one when the context calls for it. Nothing is hidden unless asked, with one exception:
cloned third-party repos are marked origin = vendor and stay out of search and lists until you
ask for them. See docs/USER_GUIDE.md.
The package is codemem-mcp; the command it installs is codemem.
Data goes to ~/.codemem/ unless CODEMEM_DB or CODEMEM_DATA says otherwise. Register it with
Claude Code:
That is the server alone. For the systemd units, the commit feed and the session hooks, use the quick start below from a clone of the repository.
No authentication. Anyone who can reach the port can read and write everything. Keep the
127.0.0.1:in-p: a bare-p 8055:8055publishes the port on every interface, and Docker's port rules bypass most host firewalls.
The image needs none of the optional pieces: embeddings are off (CODEMEM_EMBED=0) and there is no
git host. Pass -e OLLAMA_URL=... -e CODEMEM_EMBED=1 to turn semantic search on.
Server (once, on the git host):
Client (each machine, once):
Then in any Claude Code session the codemem tools are available and every session starts with a
brief of the current project (if codemem knows it) and ends with an automatic session record.
| Tool | Use |
|---|---|
search | hybrid BM25 + embedding search over everything. First call before building anything |
project_brief | everything about one project, by name or working directory |
list_projects | filter by status, tag, machine, audience, visibility |
update_project | description, purpose, status, tags, audience, origin (own/vendor), visibility, maturity |
register_asset | record a reusable script/module/prompt/skill/service with a one-line usage |
find_assets | search or list assets by kind/tag/project, or by imported library |
log_session | what was done, decided, used, abandoned, and what is next |
add_note | decision, howto, resource, issue, idea |
delete_note | remove a note written by mistake, with its index and embedding rows |
link_items | project uses / could-reuse / supersedes / derived-from another |
rate | maturity: authoritative, usable, experimental, antiquated, sunset, broken, junk, with a why |
verify | "I ran it today and it works": restarts the computed trust score's freshness clock |
handoff / list_handoffs | track one implementation replacing another through candidate, shadow, verified, promoted |
howto | how to publish, add a machine, use codemem, plus whatever docs are ingested |
activity | commits and sessions across all machines, last N days |
add_scan_root / scan_path | register a directory to scan; scan now (server machine) |
stats | counts, machines, index coverage |
help | workflow, tool list, label vocabularies. Also /codemem in Claude Code, /codemem <query> searches |
GIT_ROOT is backfilled; a post-receive hook posts each push to
/ingest/push so new commits appear within seconds. See docs/git-commit-feed.md.client/codemem_agent.py once a root is registered.auto-discovered assets, with shared-code links between repos.CODEMEM_DOC_SOURCES (this repo by default), re-hashed every sync.Storage: $CODEMEM_DB (default ~/.codemem/codemem.db, SQLite, WAL). Backups: $CODEMEM_BACKUP_DIR
(default ~/.codemem/backups/).
Extracted from a system that has been running daily against 165 projects, 735 assets and 2,000+ commits across three machines. It is stable for that use, but it has had one operator, so expect rough edges the moment your layout differs from that one. Issues and pull requests welcome; please open an issue before a large change so we can agree on the shape. See CONTRIBUTING.md for how to run it locally and the two rules that are not negotiable, and SECURITY.md for what it stores and how to report a vulnerability.
Run scripts/smoke.sh before pushing: it boots the server against a throwaway database and
checks that health, the web UI, the JSON API and the schema all come up.
Apache License 2.0. See LICENSE.
Copyright © 2026 Jeff Angelcyk.