IJFW plugin
IJFW adds shared memory, plan gates and cross-audit to Claude Code, Codex, Gemini CLI, Cursor and other coding tools
- 215 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The installer changes the configuration of Claude Code, Codex, Cursor and other tools and puts files in `~/.ijfw`
- Cross-audit sends file contents to several models through your keys and CLIs
Install
- Works in
Manual install
npm install -g @ijfw/install && ijfw-installNeeds Node 18+, Git and bash. The installer finds your tools and backs up modified files as `.bak`.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The installer changes the configuration of Claude Code, Codex, Cursor and other tools and puts files in `~/.ijfw`
- Cross-audit sends file contents to several models through your keys and CLIs
This is third-party code. Review the repository files before installing.
What it does
IJFW is a local layer over AI coding tools: one install with `ijfw-install` connects memory and workflows to 16 agents through MCP and a rules-file fallback. Decisions and project state live in plain markdown under `.ijfw/memory/`, so context survives sessions and moves from Claude Code to Cursor. Recall ranks entries by keyword match and recency, and a dream cycle promotes proven patterns and removes stale notes. The workflow follows brainstorm, plan, execute, verify, ship with gates you approve yourself. Cross-audit sends a file to several AIs from different labs in parallel, and findings are tagged consensus or contested. Runs go through your own CLIs or API keys. The package also ships a bench of specialist roles, a local cost and session dashboard through `ijfw dashboard start` and a shared design contract across agents. There is no telemetry, and the data stays on your disk.
Who it is for. Developers who work across several AI coding tools at once and want shared project memory, plans with sign-off and code checks by several models.
Good fit when
- A decision was made in Claude Code on Monday and Cursor has to know it by Wednesday
- A feature needs a staged plan with your sign-off before the agent writes code
- A file must be checked for hallucinations by three models through `cross-audit this file`
- You want a dashboard of agent costs and sessions on one timeline
Not a fit when
- You work in one agent with no shared context and its own memory is enough
- You cannot touch tool configs: the install writes entries into each one
- You want a cloud service with an account and subscription: IJFW runs locally
Example request
Recall my project and show the validation decisions we made in earlier sessionsLimitations
- The install changes the configuration of every tool it finds and puts files in `~/.ijfw`. Each modified file is backed up as `.bak.<timestamp>`.
- Installation needs Node 18 or newer, Git and a bash shell, which Git for Windows provides.
- Cross-audit runs through your own CLIs or API keys, so model costs stay with you and three labs mean three accesses.
- The memory quality figures in the README come from the project itself on the open LongMemEval test, with no independent check in the repository.
How to disable. Run `ijfw-install --uninstall` or `ijfw uninstall`. It removes IJFW entries from tool configs and the files in `~/.ijfw`, backing up every modified file as `.bak` first.
FAQ
How do I install IJFW?
Run `npm install -g @ijfw/install && ijfw-install`. It needs Node 18 or newer, Git and a bash shell. The README gives a separate PowerShell installer.
Is it only a Claude Code plugin?
No. It is richest in Claude Code, but the same memory and workflows reach 16 tools through MCP and a rules-file fallback.
Where is the memory stored?
In plain markdown files under `.ijfw/memory/` in your repository, with decision receipts as JSON on disk. There is no cloud account and no telemetry.
How does `cross-audit` work?
Three AIs from different labs review the file in parallel and findings are tagged consensus or contested. Requests go through your own CLIs or API keys.
How do I remove IJFW without losing configs?
`ijfw uninstall` removes the IJFW entries and files under `~/.ijfw` and backs up every modified file as `.bak.<timestamp>`. Other plugins and MCP servers stay untouched.
Discussion
No questions or reviews yet.
Related
PluginsLocal memory for AI agents in plain Markdown files: it keeps only what you approve and serves it through MCP
The lwc CLI in Rust: persistent agent memory as a source-grounded wiki on SQLite, for Claude Code, Codex, Cursor and more
One assistant with shared memory and a dashboard on top of Claude Code, Codex, Cursor, Copilot and other AI CLIs you already have
A plugin for Claude Code and Codex: a shared task wall, project memory, reports and a dashboard that shows agents' work across sessions
The ee CLI gives agents durable local memory: rules, decisions and anti-patterns with scores and provenance
Shared infrastructure for Claude Code, Codex, Antigravity CLI and OpenCode: a task lifecycle, skills and sandboxes.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| IJFW (this card) | no | 215 | ||
| agentmemory | yes | not required | 30,340 | an MCP server with memory only, no plans or audit |
| MCP Memory Service | no | not required | 2,117 | a memory service for agents, no workflow or cross-checking |
README in short
The README presents IJFW as one shared brain for AI tools and walks through ten blocks: memory, build discipline, cross-audit, specialists, token spend, a dashboard, personalization, a design contract, 16 supported tools and data ownership. Then come the quickstart, the agent list, a what-this-is-not section, uninstall and FAQ.