Add config toggles for orchestrator skills and MCP#28942
Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 1982f10023
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 34d0d4908d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Why
Orchestrator-provided skills and Codex Apps MCP tools add model-visible instructions, resources, and tools beyond the local workspace. Hosts need config-level switches to disable those orchestrator-owned surfaces independently, without disabling regular skills or regular MCP servers.
What changed
[orchestrator.skills].enabledand[orchestrator.mcp].enabledconfig entries, both defaulting totrue.config.schema.jsonand in the config lock so resolved thread configuration preserves the same orchestrator exposure decisions.orchestrator.skills.enabledthrough the app-server skills extension so disabled orchestrator skills do not expose theskillsnamespace or inject orchestrator skill context.orchestrator.mcp.enabledwhile leaving non-Codex-Apps MCP tools available.Verification
orchestrator.mcp.enabled = falseremoves Codex Apps tools while preserving regular MCP tools.orchestrator.skills.enabled = falseprevents orchestrator skill tools, prompts, and resource reads from reaching the model turn.