Provides integration with OpenAI's Codex CLI, enabling capabilities such as code execution, automated code writing, and parallel code reviews with session management and real-time monitoring.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@mcp-codex-dev Review my current code changes for quality and logic errors"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP Server for integrating Codex CLI into Claude Code workflows with session management and real-time progress monitoring.
Prerequisites: Codex CLI installed and configured.
In .mcp.json file
{
"mcpServers": {
"mcp-codex-dev": {
"command": "npx",
"args": ["-y", "mcp-codex-dev"]
}
}
}or
claude mcp add mcp-codex-dev -- npx -y mcp-codex-dev Windows OS:
{
"mcpServers": {
"mcp-codex-dev": {
"command": "cmd",
"args": ["/c", "npx", "-y", "mcp-codex-dev"]
}
}
}or
claude mcp add mcp-codex-dev -- cmd /c npx -y mcp-codex-devRelated MCP server: codex-bridge
Tool | Description |
| Clean dialogue without templates, supports session resume |
| Codex CLI with built-in test-driven development prompt template |
| Code review (spec + quality in parallel), supports resume |
| Environment and config diagnostics |
| List tracked sessions |
| Discard sessions |
Create ~/.mcp/mcp-codex-dev/config.json:
{
"model": "gpt-5.2",
"sandbox": "danger-full-access",
"timeout": 300000,
"tools": {
"tdd": { "model": "gpt-5.3-codex", "sandbox": "danger-full-access", "timeout": 2000000},
"review": { "model": "gpt-5.2", "sandbox": "danger-full-access", "timeout": 3000000},
"health": { "enabled": false }
}
}Top-level model / sandbox / timeout are global defaults. The tools section overrides per tool (keyed by tool name). Set "enabled": false to disable a tool.
Per-project config can be placed at <project>/.mcp/mcp-codex-dev.config.json.
A local HTTP server starts at http://localhost:23120 showing real-time progress. Configurable via progressPort.
When multiple MCP server instances run simultaneously, they share a single progress page: the first instance binds the port and serves the UI, and other instances forward their progress events to it.
Session tracking metadata is stored per project at <project>/.mcp/mcp-codex-dev/sessions.json (where <project> is the Git repo root if applicable).
Codex CLI's own session files remain in ~/.codex/sessions/<id>/.
No license.
Invoke Codex CLI with a plain instruction. No template, no context wrapping — just raw execution. Supports session resume.
Inputs: instruction (required), sessionId, workingDirectory.
Run environment and configuration checks (Codex CLI, config, Git, filesystem).
Inputs: workingDirectory, checkGit, ensureTrackingDir.
Invoke Codex CLI for code review. Uses built-in code-reviewer template. Pass git range and description info, returns structured review results. Supports continuing previous review sessions via sessionId. In 'full' mode, returns specSessionId and qualitySessionId separately (use with reviewType 'spec'/'quality' to resume each).
Inputs: instruction (required), whatWasImplemented (required), baseSha (required), headSha, +7 more.
Discard Codex sessions. By default, refuses to discard sessions marked active unless force=true.
Inputs: sessionIds (required), force, workingDirectory.
List tracked Codex sessions. Can filter by type and status.
Inputs: type, status, workingDirectory.
Invoke Codex CLI to implement code using strict Test-Driven Development (Red-Green-Refactor). Injects TDD methodology template that enforces writing tests before production code. Supports session resume for iterative TDD cycles.
Inputs: instruction (required), sessionId, workingDirectory, planReference, +2 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 6 tools
Each tool has a clearly distinct purpose: exec for raw CLI execution, health for environment checks, review for code reviews, session_discard and session_list for session management, and tdd for TDD-based implementation. There is no overlap or ambiguity in functionality, making tool selection straightforward for an agent.
Most tools follow a consistent verb-based naming pattern (exec, health, review, tdd), but session_discard and session_list deviate with a noun_verb structure. This minor inconsistency is still readable and does not severely impact usability, though it breaks the overall pattern slightly.
With 6 tools, the server is well-scoped for its purpose of interacting with Codex CLI. Each tool serves a specific, necessary function (execution, health checks, reviews, session management, TDD), and none feel redundant or out of place, making the count appropriate.
The tool set covers core workflows for Codex CLI interaction, including execution, health, reviews, and session management. However, there are minor gaps, such as no explicit tool for creating or updating sessions beyond resume capabilities, which agents might need to work around using exec or other tools.
Shared memory MCP server for Claude Code, Codex, Gemini CLI and Cursor. Cited, editable.
Augments MCP Server - A comprehensive framework documentation provider for Claude Code
FixMyPrompt is a remote MCP server for coding agents. It saves the steps your agent worked out as project memories and fetches them in later sessions, scores and rewrites prompts, and checks CLAUDE.md and AGENTS.md files. Works with Claude Code, Codex and Cursor. Free plan available.
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
