A lightweight, zero-dependency scheduler that lets command-line AI agents like Claude Code and Codex work in the background and ask you for decisions via cards only when needed.
im4codes/imcodes
An IM-style control layer for AI coding agents, adding shared memory, remote terminal and desktop access, managed MCP tools, and cross-agent audit.
- Stars
- 960
- Forks
- 121
- Open issues
- 1
- Last commit
- 18 hours ago
What is it?
IM.codes connects AI coding agents such as Claude Code, Codex, Gemini CLI, and OpenClaw to the machines where they run. A daemon on each dev machine manages agent sessions through tmux or SDK transports, and a self-hosted server relays access to a web or mobile dashboard. From there you can open terminals, browse files and git changes, preview localhost, and get notified when work finishes.
The project also adds a shared memory layer that turns completed assistant outputs into summaries recalled in later sessions, and a managed MCP tool surface for memory search, agent-to-agent messaging via `imcodes send`, and cron scheduling. Multi-agent Team discussions let several providers review work before code is written, and a supervisor can auto-continue, hand back, or run an audit-rework loop per session.
IM.codes is a personal MIT-licensed project installed with npm. It is self-hostable with Docker and PostgreSQL, requires Node.js 22 or later, and supports macOS and Linux, with Windows listed as experimental.
Key features
- 01Remote terminal and file/git views in any browser, without SSH or VPN.
- 02Shared memory that recalls summarized past work across sessions and providers.
- 03Managed MCP tools for memory search, agent messaging, and cron scheduling.
- 04Team discussions where multiple providers review work before implementation.
- 05Per-session supervisor that auto-continues, hands back, or runs audit loops.
- 06Localhost preview tunneled through the server, with HMR support.
Use cases
- Checking on and steering a long-running coding agent from a phone while away from the desk.
- Having one agent ask another to review changes or run tests via `imcodes send`.
- Triggering agent work from CI failures, webhooks, or log monitoring scripts.
- Running a multi-model review of a spec before implementation begins.
Quick start
- 1
Download the app from the App Store or use the web app at app.im.codes
Things to know
- The README states it is an actively developed personal project with no warranties, SLA, or stability guarantees.
- Shared enterprise context is still under active development and has not been fully production-tested.
- Requires a supported coding agent provider and a self-hosted or account-based setup to use the full feature set.
In collections
Topics
FAQ
- What is imcodes?
- An IM-style control layer for AI coding agents, adding shared memory, remote terminal and desktop access, managed MCP tools, and cross-agent audit. IM.codes connects AI coding agents such as Claude Code, Codex, Gemini CLI, and OpenClaw to the machines where they run. A daemon on each dev machine manages agent sessions through tmux or SDK transports, and a self-hosted server relays access to a web or mobile dashboard. From there you can open terminals, browse files and git changes, preview localhost, and get notified when work finishes.
- Is imcodes free to use? What license does it have?
- imcodes is open source under the MIT license. Check the license terms before using it in commercial products.
- Is imcodes still maintained?
- Actively maintained. The last commit to imcodes was on Nov 7, 2026.
- What can I use imcodes for?
- Checking on and steering a long-running coding agent from a phone while away from the desk. Having one agent ask another to review changes or run tests via `imcodes send`. Triggering agent work from CI failures, webhooks, or log monitoring scripts. Running a multi-model review of a spec before implementation begins.
- How do I install imcodes?
- Download the app from the App Store or use the web app at app.im.codes
- Where can I get imcodes?
- The source code and installation instructions are on GitHub: https://github.com/im4codes/imcodes
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Long-term memory for coding agents that stores memories in a local SQLite database and shares them across Claude Code, Codex, Cursor and other MCP clients.
Local-first, governed memory layer that gives AI agents like Claude Code, Cursor and Codex persistent recall that admits when it doesn't know.
A free, two-way Figma MCP server that lets coding agents turn Figma designs into code and write designs back to the canvas.
Self-hosted personal AI assistant that builds a memory graph from your email, meetings, and code tools, then plans and runs tasks for you.
Open-source desktop and self-hosted chat app where each bot is a local Claude, Codex or Grok agent with its own model, cloud computer and connected apps.