Skip to content

release: 0.6.0 behavior-suite results and plan-only checkpoint fix - #7

Merged
sofus-common merged 2 commits into
mainfrom
fix/plan-only-checkpoint-approval
Oct 10, 2026
Merged

sofus-common merged 2 commits into
mainfrom
fix/plan-only-checkpoint-approval

Conversation

@sofus-common

Copy link
Copy Markdown
Contributor

Summary

  • Fix: the router's plan-only clause now lists the Controlled approval after checkpoint 1 among the approvals still needed. The 0.6.0 behavior run on Claude Code (claude-sonnet-5-5) failed payment because that approval was not named; ai-native-plan, which states the rule, was skipped.
  • Dates the 0.6.0 changelog entry with the behavior-suite results.

Verification

  • Behavior suite (12 cases, independent transcript review): Claude Code 2.1.296 claude-sonnet-5-5 11/12, Codex 0.161.0 gpt-6.1-sol 12/12. Filesystem checks and executable acceptance passed on both.
  • Paired payment runs on claude-sonnet-5-5: gap in 1/3 on 0.5.1, 3-4/4 on 0.6.0 before the fix, 1-2/4 after. Codex after fix: 2/2 name the approval. Small samples, direction only.
  • check_version.py (+ self-test), test_behavior_check.py, test_edit_guard.py pass.
  • Run limits are recorded in the changelog.

🤖 Generated with Claude Code

sofus-common and others added 2 commits October 10, 2026 14:56
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@sofus-common
sofus-common merged commit 446cdff into main Oct 10, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant