fix: require the per-wake Codex MCP connection - #22
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A Codex turn could continue with only built-in tools when its per-wake MCP connection failed, leaving required capabilities unavailable without failing the turn. Every Codex launch now explicitly enables and requires the Daimon MCP server, so startup failures stop before model work and follow the existing redacted error path.
Covers standalone and strict sandbox launches without changing scheduling, retry behavior, public contracts, or successful tool calls.
Validation: 633 tests passed (4 platform skips), typecheck, production and test-runtime builds, contract/native/production-closure checks, and boundary checks. Independent review passed. Removing either required setting fails all four new regression cases. An isolated probe of the pinned Codex CLI, using both rendered launch modes and a local mock provider, confirmed failed MCP startup makes zero model requests; the prior optional setting produced a normal completed turn. No runtime deployment is included.