# Codex MCP > Use this `codex-mcp` skill to instruct another AI Agent to invoke a tool or execute a workflow on your behalf for the following scenario, 1.When you are requested to invoke a mcp tool that you don't have access (can use `ListMcpResourcesTool` to check) such as `context7`, `github`, `linear`, `jira`; 2.When you are requested to execute a complex workflow that require interacting with multiple mcp tools and bash CLI commands for example `github-project-create-items`, `github-project-list-items` workflows; - Skill: `gosu-code/codex-mcp` (Agent Skill, multi-file: 5 files) - Install (CLI): `npx skillmds@latest add gosu-code/codex-mcp` - Raw SKILL.md: https://api.skillmd.com/api/skills/gosu-code/codex-mcp/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: gosu-code (https://skillmd.com/u/gosu-code) - Updated: 2026-09-10 - Page: https://skillmd.com/skills/gosu-code/codex-mcp --- # codex-mcp You MUST check whether `codex` CLI and MCP server are available in the current bash environment with this bash command: `codex mcp list`. If not available, run the following bash command to install the `codex` CLI: `npm install -g @openai/codex`. Then inform user to perform authentication with an OpenAI account by running `codex login`. Ignore this instruction if the CLI and MCP server are already available. ## Tools ### mcp__codex__codex Run a Codex session. Accepts configuration parameters matching the Codex Config struct. Supported properties: | Property | Type | Description | | ----------------------- | ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------ | | **`prompt`** (required) | string | The initial user prompt to start the Codex conversation. | | `approval-policy` | string | Approval policy for generated shell commands: `untrusted`, `on-failure`, `on-request`, `never`. | | `base-instructions` | string | Overrides the default base instructions for the session. | | `compact-prompt` | string | Prompt used when compacting the conversation. | | `config` | object | Path to individual config settings toml file that override `$CODEX_HOME/config.toml` (see [codex-config-toml-structure.md](references/codex-config-toml-structure.md) to understand how to create this file). | | `cwd` | string | Working directory for the session. Relative paths resolve from the server process root. | | `developer-instructions`| string | Developer instructions that should be injected as a developer role message. | | `model` | string | Optional override for the model name (e.g., `gpt-5.5`, `gpt-5.4-mini`, `gpt-5.4`, `gpt-5.4-nano`, `gpt-5.2`, `gpt-5.2-codex`) default is `gpt-5.4`. | | `profile` | string | Name of the profile providing default options (defined in `config.toml`). | | `sandbox` | string | Sandbox mode: `read-only`, `workspace-write`, or `danger-full-access`. | **Response:** | Property | Type | Description | | ------------------------- | ------ | -------------------------------------------------------- | | **`threadId`** (required) | string | Thread identifier (UUID) for the Codex session. | | **`content`** (required) | string | The response content from the Codex session. | ### mcp__codex__codex-reply Continue a Codex session by providing the conversation id and follow-up prompt. Supported properties: | Property | Type | Description | | ------------------------------- | ------ | -------------------------------------------------------- | | **`prompt`** (required) | string | Next user prompt to continue the Codex conversation. | | **`threadId`** (required) | string | Identifier of the thread to continue. Must be a valid UUID (e.g., `019c08dc-ca4c-7a21-a0ce-b783c2cdec56`). | | `conversationId` | string | **Deprecated.** Use `threadId` instead. | **Response:** | Property | Type | Description | | ------------------------- | ------ | -------------------------------------------------------- | | **`threadId`** (required) | string | Thread identifier (UUID) for the Codex session. | | **`content`** (required) | string | The response content from the Codex session. | ## Operational Requirements - Always route qualifying requests through `mcp__codex__codex` with `sandbox` set to `danger-full-access` (MUST) , `approval-policy` set to `never` (MUST) and `model` set to `gpt-5.4-mini` (SHOULD). - When an unknown MCP tool name is requested, use `ListMcpResourcesTool` to confirm if it is available. If not available, delegate to `codex` to invoke this MCP tool by crafting a `prompt` with all necessary information and file references. - When you need to run a command that would emit large stdout output but only a subset of that output is needed, MUST provide a precise `prompt` when delegating via `mcp__codex__codex` so the downstream agent knows which portion of the large output to collect or which external tool to execute. - Prefer passing context via a local file when delegating: include a relative path prefixed with `@` inside the `prompt` (example prompt: "use mcp Github to submit all review feedback comments in this file @ghpr-code-review/pr_47_feat_e2e_tests_review_feedback.md Make sure to include correct line number for each comment"). - When `mcp__codex__codex` returns a result requiring your follow-up, continue the conversation with `mcp__codex__codex-reply`, reusing the `threadId` present in the previous response (must be a valid UUID). ## Image Generation When the user requests to generate an image, delegate to `mcp__codex__codex` using `model` set to `gpt-5.5` (REQUIRED for image generation). The downstream Codex agent must: - Copy the generated image file to the current working directory (`cwd`). - Return the **absolute path** of the copied image file in the response. When crafting the `prompt`, include explicit instructions such as: > "Generate an image of `{description}`. > Must use built-in `image_gen` tool to generate image. > Read the imagegen system skill at ~/.codex/skills/.system/imagegen/SKILL.md to understand how to use the `image_gen` tool. > Copy the output image to the current working directory and respond with the absolute path of the copied image file." Example delegation parameters: - `model`: `gpt-5.5` - `cwd`: the user's current working directory (absolute path) - `sandbox`: `danger-full-access` - `approval-policy`: `never` - `prompt`: e.g. `"Generate a PNG image of a red origami fox sitting on a stack of books in a cozy library, soft warm lighting, shallow depth of field. Must use built-in `image_gen` tool to generate image. Read the imagegen system skill at ~/.codex/skills/.system/imagegen/SKILL.md to understand how to use the `image_gen` tool. Copy the output to the current working directory, and respond with ONLY the absolute path of the copied image file (no extra commentary)."` After receiving the response, surface the absolute path of the generated image to the user. ## Known MCP Server Tools - When `github` MCP server tool (`mcp__github`, `mcp__github__*`) is requested, see [mcp-github-usage.md](references/mcp-github-usage.md) for detailed usage instructions. Use the information in this file to craft a precise `prompt` when delegating via `mcp__codex__codex` so the downstream agent can use the `github` MCP server tool exactly as intended. - When `linear` MCP server tool (`mcp__linear`, `mcp__linear__*`) is requested, see [mcp-linear-usage.md](references/mcp-linear-usage.md) for detailed usage instructions. Use the guidance in this file to compose precise prompts when delegating via `mcp__codex__codex` so the downstream agent invokes Linear tools correctly. - When `context7` MCP server tool is requested, see [mcp-context7-usage.md](references/mcp-context7-usage.md) for detailed usage instructions.