AgentScope validates trace structure today. The next layer should flag suspicious or incomplete traces.
Possible checks:
- Run has edits but no verification command
- Run has failed tests but no later passed test
- High-risk edits have no evidence notes
- Summary exists but no changed files are recorded
Expected output:
- CLI warnings
- Optional PR summary checklist
- Documentation for quality rules
Non-goals:
- Blocking all trace imports
- Inferring hidden agent reasoning
AgentScope validates trace structure today. The next layer should flag suspicious or incomplete traces.
Possible checks:
Expected output:
Non-goals: