It would be very helpful if Harness printed a small startup summary for each new session showing what prompt/skill context was actually loaded.
Something like:
- AGENTS.md loaded from:
- Skills loaded from:
- /path/to/skill-a/SKILL.md
- /path/to/skill-b/SKILL.md
This would make debugging much easier when a skill or prompt is expected to be available but the model behaves as if it isn't.
Right now it's hard to tell whether the problem is:
- the file was not discovered
- the file was discovered but not injected
- or the model just ignored available context
A simple startup display would make this deterministic and much easier to verify.
It would be very helpful if Harness printed a small startup summary for each new session showing what prompt/skill context was actually loaded.
Something like:
This would make debugging much easier when a skill or prompt is expected to be available but the model behaves as if it isn't.
Right now it's hard to tell whether the problem is:
A simple startup display would make this deterministic and much easier to verify.