Repository navigation
Add Codex and Claude subscription runners - #2
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Grudges can now replay reviewed tasks using the installed Codex or Claude CLI and its existing subscription login.
--runner local-codexand--runner local-clauderun sequentially, preserve native credentials, and grade candidates offline in Docker. API replay through Harbor remains the default.Each task must pass baseline/reference controls before consuming allowance. Rules comparisons use fresh workspaces, paired attempt ordering and frozen inputs. Restricted Claude receives the root
CLAUDE.mdexplicitly after live canaries showed native auto-loading was disabled. Receipts preserve errors and missing attempts; subscription cost is never reported as zero.Validated on Linux: 78 tests passed on Python 3.11/3.13 (one optional test skipped), all 28 bundled baseline/reference controls passed, and both live subscription runners passed a two-setup comparison with a marker present only in the rules arm. Wheel/sdist, installed-wheel suite, secret scan and desktop/mobile cards checked.
Local mode requires
--trust-local: agents retain host access, and Claude's file-only tools differ from Codex's shell. It is not an isolated benchmark. Prepares GitHub prerelease 0.2.0a1.