About Agent Config Bridge
Agent Config Bridge is a local MCP server for moving instruction files and MCP settings among Claude Code, Codex CLI, Cursor, and Gemini CLI. It scans configurations, converts formats, and prepares migrations without network calls; file changes require an explicit write option.
Key features
- 01Convert instruction files between agents
- 02Convert MCP configuration formats
- 03Scan project configs and skills
- 04Generate migrations with dry run
- 05Check documented config locations
Use cases
- 01Move project instructions to another coding agent
- 02Convert MCP settings between supported formats
- 03Preview missing configurations before migration
MCP support
Runs locally over stdio and exposes tools for detecting, converting, and migrating coding-agent configurations. Migration writes are opt-in through `migrate_project` with `write: true`.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEagent-config-bridge
An MCP server that moves your coding-agent setup between Claude Code, Codex CLI, Cursor and Gemini CLI: instruction files (CLAUDE.md, AGENTS.md, .cursor/rules/*.mdc, .cursorrules, GEMINI.md) and MCP server configs (mcpServers JSON ⇄ Codex config.toml ⇄ Gemini settings.json).
Runs locally over stdio. No network calls, no telemetry. It writes files only when you call migrate_project with write: true.
Prefer a web UI? Use the same converter in your browser at getaicodingagent.com.
Tools
| Tool | What it does |
|---|---|
convert_rules | Instruction text from one tool → files and paths for the others. Can also share one AGENTS.md through a symlink or an @AGENTS.md import |
convert_mcp_config | MCP servers: Claude Code / Claude Desktop / Cursor JSON ⇄ Codex [mcp_servers.*] TOML ⇄ Gemini httpUrl/url |
detect_agent_configs | Scans a project read-only for every agent config file and skills folder, and lists which tools have no config yet |
migrate_project | Generates the missing configs from the existing ones. Dry run by default. Existing files are skipped unless overwrite: true; .gemini/settings.json is merged |
compat_matrix | Where each tool reads its config, with official doc links and a verified date. Anything the official docs don't state is returned as unknown |
Install
You need Node.js 18 or later and npm 10 or later (npm 9.2 fails to run GitHub packages through npx). The package is not on npm. Run it straight from GitHub (dist/ is committed, so there's no build step):
npx -y github:onehumanbeing/agent-config-bridge
Claude Code
claude mcp add agent-config-bridge -- npx -y github:onehumanbeing/agent-config-bridge
Codex CLI
codex mcp add agent-config-bridge -- npx -y github:onehumanbeing/agent-config-bridge
or add this to ~/.codex/config.toml:
[mcp_servers.agent-config-bridge]
command = "npx"
args = ["-y", "github:onehumanbeing/agent-config-bridge"]
Cursor: .cursor/mcp.json (project) or ~/.cursor/mcp.json (global). Gemini CLI: .gemini/settings.json or ~/.gemini/settings.json:
{ "mcpServers": { "agent-config-bridge": { "command": "npx", "args": ["-y", "github:onehumanbeing/agent-config-bridge"] } } }
MCPB bundle: download agent-config-bridge-0.1.0.mcpb from the v0.1.0 release and open it in an MCPB-compatible client (for example, double-click it to install in Claude Desktop). Dependencies are bundled.
From a clone
git clone https://github.com/onehumanbeing/agent-config-bridge && cd agent-config-bridge
claude mcp add agent-config-bridge -- node "$PWD/dist/index.js"
MCP Registry name: io.github.onehumanbeing/agent-config-bridge
Example prompts
- "Detect the agent configs in this repo and migrate them to Codex and Cursor (dry run first)."
- "Convert this Claude Desktop config to Codex TOML."
- "Where does Gemini CLI read its MCP config? Use compat_matrix."
Links
- Web converter and compatibility table: https://getaicodingagent.com
- Config matrix for 17 agents: https://getaicodingagent.com/config/
- Compare coding agents: https://getaicodingagent.com/compare
Develop
npm install && npm test # builds, then runs unit tests plus an MCP client smoke test
MIT © onehumanbeing
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.