Docmancer: one shared memory for coding agents
Docmancer
A CLI that indexes memory, rules and sessions of Claude Code, Codex, Cursor and Gemini and hands durable decisions to every agent
- 133 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Reads session history, rules and memory of all agents on disk, which can hold code and secrets
- Claude Code and Codex hooks fire automatically and add findings to the context
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/docmancerDetects 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.
- Reads session history, rules and memory of all agents on disk, which can hold code and secrets
- Claude Code and Codex hooks fire automatically and add findings to the context
This is third-party code. Review the repository files before installing.
What it does
Docmancer finds on your computer what coding agents wrote down: `CLAUDE.md` and `AGENTS.md` files, Cursor rules, memory directories and session history. From this it builds a shared memory in plain Markdown files you can read, edit and commit, and serves it to Claude Code, Codex, Cursor, Cline, Gemini CLI, OpenCode and GitHub Copilot. The `docmancer setup` command finds agents, shows a plan and waits for confirmation. After it, `docmancer ask "Why did we choose Railway?"` answers with citations from other sessions, and `docmancer web` opens an interface with the shared memory tree. `docmancer consolidate` proposes durable decisions with a link to the dialogue fragment and applies nothing without your approval. For Claude Code and Codex there are hooks that pull in recall and capture sessions, and for the others a skill file is installed. `docmancer backup` and `docmancer restore` make an encrypted archive for backup and moving to another machine.
Who it is for. Developers who run several coding agents at once and are tired of re-explaining the project layout and past decisions to each.
Good fit when
- Claude Code knows why you left Railway but Codex and Cursor do not, and you want one shared memory
- Cursor still follows a rule you overturned two months ago and you need to find the contradictions
- You are moving to a new laptop and want to carry agent sessions and settings in an encrypted archive
- You want an answer with citations of which agent made a given decision and when
Not a fit when
- You work with one agent and its own memory is enough
- You need a cloud memory service for a team: paid sync is for personal devices only, while the core is local
- You are not ready to let a tool read agent session history and rules on disk
Example request
Ask docmancer why we chose Railway and show the citations from Claude Code and Codex sessions.Limitations
The core is local: SQLite FTS5, sqlite-vec and a small bundled embedding model, with no telemetry, account or key. The agent backup and restore commands are called a developer preview, and managed cloud snapshots are gated on migration and recovery checks.
Paid personal sync encrypts data and includes a 30-day trial. Automatic recall and session capture exist only for Claude Code and Codex. The last commit is from August 2026.
How to disable. Remove integrations with the `docmancer agent` commands for the agent in question, or delete skill and hook files by hand, then remove the package with `pipx uninstall docmancer`. Shared memory sits in ordinary directories `~/.docmancer/tree/` and `.docmancer/tree/`.
FAQ
How do I install Docmancer?
With `pipx install docmancer`, then `pipx ensurepath` and `docmancer setup`. `uv tool install docmancer` works too. The wizard finds agents and waits for your confirmation.
Does my data go out to the network?
No, by default: the core uses SQLite and a small bundled model, with no telemetry, account or key. The network is needed only for the optional paid sync.
How do I connect Docmancer through MCP?
With `docmancer mcp install codex`, and `docmancer mcp doctor` checks it. Claude Desktop has MCP while the skill file is uploaded manually.
Can an agent change memory without my knowledge?
No. On a request to remember, Docmancer prepares a whole file and shows the diff, the default answer is no, and a stored proposal is not applied later by a "yes" in conversation.
Is Docmancer paid?
The core is free and works on one machine. Only the optional personal sync with encryption and revision history is paid, with a 30-day trial.
Discussion
No questions or reviews yet.
Related
CLI toolsA local workbench for Claude Code, Codex, Pi and OpenCode: memory changes wait for your review, then an audit follows each turn
daidocs keeps assistant memory as plain .dai files on your disk and connects it over MCP to Claude Code, Cursor and Codex.
agentcairn keeps Claude Code, Codex and Cursor memory in Markdown files for Obsidian and hands an agent only what it needs
Local memory for Claude Code, Codex and Claude Desktop: one Markdown folder and one MCP server shared by all.
Long-term memory for agents: it distills knowledge from session history and brings it back in new tasks, one store shared across Claude Code, Codex, Cursor and
Persistent memory for coding agents: insights are stored as markdown cards and recalled at the start of the next session.
README in short
The README opens with the shared-memory problem, then shows backup and restore, install through pipx, the first minute of use, the memory directory layout, a table of supported agents and the MCP mode. A separate section covers paid sync.