Repository navigation
Test real CLI conversation recovery across same-directory splits - #1068
Merged
Merged
Conversation
Kuddev
changed the base branch from
test/agent-cold-restore-matrix-1120
to
main
November 20, 2026 11:19
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Result
The native recovery runner opens two copies of the same CLI in a shared working directory and cold-restarts the application three times. Each restored pane must accept a new message in its own original conversation. A live test entry is registered for every supported CLI, including both Cursor entry points, Antigravity, and the CodeBuddy aliases.
Design
A loopback model provides offline turns for Claude Code, Pi, OMP, Grok Build and CodeBuddy. The runner cross-checks each saved ID against the tool's native transcript and checks the model request for the correct pane history. For providers that require sign-in, a separate runner uses a dedicated test account and independent conversation challenges. Ordinary pull-request tests never contact a paid model.
Evidence