cc-codex-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CODEX_BRIDGE_HOME | No | Isolated CODEX_HOME. Refused if it resolves to the main home. | ~/.codex-reviews |
| CODEX_BRIDGE_CODEX_PATH | No | Explicit Codex executable path (.exe, .js, or an npm .cmd/.ps1 shim to unwrap). | |
| CODEX_BRIDGE_COPY_CONFIG | No | Set to 1 to copy ~/.codex/config.toml once if the isolated home has none. | |
| CODEX_BRIDGE_MCP_EXCLUDE | No | Comma-separated MCP server names not to sync. | |
| CODEX_BRIDGE_SOURCE_HOME | No | Main Codex home, used as the source of truth for config and sandbox state. | ~/.codex |
| CODEX_BRIDGE_SYNC_DESKTOP_ONLY | No | Set to 1 to also sync desktop-only servers such as node_repl. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| codex_askA | Get an independent opinion from a different model (OpenAI Codex), e.g. a read-only review of changes you made. Returns a thread_id; keep it if you may want to ask follow-ups. Codex runs as an agent in |
| codex_followupA | Continue an existing Codex thread by thread_id: re-check fixes, ask why it flagged something. Always use the thread_id from your own earlier codex_ask; never guess one. The thread keeps its original working directory and sandbox mode. Only one follow-up can run on a thread at a time; a concurrent one waits (up to timeout_sec) and then fails with "thread busy". Returns JSON: { thread_id, answer, duration_sec, warnings[], activity?, answer_file?, error? }. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 2 tools
codex_ask starts a new independent-review thread, while codex_followup continues an existing one via thread_id. The two purposes are cleanly separated with no overlapping scope, so an agent can choose reliably.
Both tools use the same codex_ prefix with snake_case action suffixes (ask, followup). The pattern is predictable and no conventions are mixed.
Two tools is lean but well-matched to a narrow purpose: initiate an opinion thread and extend it. It is slightly below the typical 3-15 range, but each tool clearly earns its place.
The ask/follow-up pair covers the core lifecycle of getting and continuing an independent review. Minor gaps such as listing or cancelling threads exist but are not required for the stated workflow.