Remnic CLI
Shared memory for Claude Code, Codex, Cursor and other agents in Markdown files on disk, with retrieval that explains its results
- 232 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The daemon stores and injects the content of your conversations with agents and sends texts to a model provider for extraction.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/remnicDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The daemon stores and injects the content of your conversations with agents and sends texts to a model provider for extraction.
This is third-party code. Review the repository files before installing.
What it does
Remnic is local memory and context for AI agents. Records are plain Markdown files with YAML frontmatter on your disk, with local JSON and SQLite state beside them and no external database needed. You can read them with `cat` and `grep`, edit and version them. One memory serves every tool: a preference told to one agent is known to the others. The `remnic` command installs from npm. `remnic init` writes `remnic.config.json`, `remnic daemon start` launches the background server, `remnic status` shows state, and `remnic query "hello" --explain` breaks down which retrieval tier produced a result and why. `remnic doctor` checks the setup. Search is hybrid: BM25, vectors and reranking over rebuildable indexes with graph recall, record worth scoring and per-result provenance. Remnic watches conversations, extracts durable knowledge and injects it back when needed. Integrations: OpenClaw through a plugin, Claude Code and Codex CLI through hooks and MCP, Cursor, Hermes, Replit, Pi and any MCP client.
Who it is for. Developers who work with several agents at once and want preferences and knowledge to carry between them without retelling.
Good fit when
- You tell one agent a preference and want Claude Code, Codex and Cursor to know it too.
- Memory must live in files readable and editable with ordinary tools.
- You need to understand why this record was found: `remnic query --explain` fits.
- You want to connect memory to any MCP client over HTTP or stdio.
Not a fit when
- You do not want a background server: Remnic runs a daemon started with `remnic daemon start`.
- Node.js 22.12 is unavailable: it is the minimum version.
- You need memory in a shared cloud service: Remnic is built for local storage.
Example request
Install Remnic, run `remnic init`, start the daemon and check `remnic query "hello" --explain`, then connect it to my agent.Limitations
Node.js 22.12 or newer is required. Knowledge extraction needs a model: an OpenAI key, the OpenClaw gateway model chain or a local LLM, while retrieval-only mode needs none. QMD gives the best hybrid search, without it Remnic falls back to embedding search and then to recency-ordered reads. The daemon is protected by the `REMNIC_AUTH_TOKEN` token set before launch. The native OpenClaw plugin needs a write to `openclaw.json` and a gateway restart, and without the memory slot hooks do not fire. The old `engram` name stays as a transitional forwarder and temporarily appears in log prefixes.
How to disable. Stop the daemon with `remnic daemon stop`, remove the package with `npm uninstall -g @remnic/cli` and, if you wish, the memory directory and `remnic.config.json`.
FAQ
How do I install Remnic?
Run `npm install -g @remnic/cli`, then `remnic init`, set `REMNIC_AUTH_TOKEN` and start `remnic daemon start`. Node.js 22.12 or newer is required.
Where is memory stored?
In plain Markdown files with YAML frontmatter on your disk, plus local JSON and SQLite. No separate database service is needed.
Which agents does it work with?
OpenClaw, Claude Code, Codex CLI, Cursor, ChatGPT developer mode, Hermes, Replit, Pi, Oh My Pi, Factory Droid and any MCP client.
How do I check it works?
With `remnic doctor`, and for OpenClaw also the `gateway_start fired` line in `~/.openclaw/logs/gateway.log`.
Can I run without an OpenAI key?
Yes, by routing extraction to a local LLM. Retrieval-only mode needs no model at all.
Discussion
No questions or reviews yet.
Related
CLI toolsReMe keeps agent memory in Markdown files with wikilinks and installs as a plugin for Claude Code, Codex and Qoder
Local memory for AI agents in plain Markdown files: it keeps only what you approve and serves it through MCP
Shared local memory for AI agents: Claude Code, Codex, Cursor and others find the agreements of past sessions in it and record new ones
Persistent memory for coding agents: insights are stored as markdown cards and recalled at the start of the next session.
Local memory for Claude Code, Codex and Claude Desktop: one Markdown folder and one MCP server shared by all.
Persistent agent memory in a folder of linked markdown notes, so Claude Code remembers a project across sessions
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Remnic (this card) | no | 232 | ||
| agentmemory | yes | not required | 30,340 | agentmemory provides memory through MCP, while Remnic stores it in Markdown and connects through hooks and MCP to about ten agents. |
| agent-memory | no | 4,315 | agent-memory keeps notes of one agent, while Remnic gives a shared memory with hybrid search for all agents. |
README in short
The README covers local storage, a quick start for any agent and for OpenClaw, an integration table with registration commands and the family of the author neighboring projects. MIT license, the `@remnic/cli` package is on npm.