Skip to content

Add audit-agent-skill#1

Closed
realmarcin wants to merge 2 commits into
berkeleybop:mainfrom
realmarcin:main
Closed

Add audit-agent-skill#1
realmarcin wants to merge 2 commits into
berkeleybop:mainfrom
realmarcin:main

Conversation

@realmarcin

Copy link
Copy Markdown

Summary

  • Adds audit-agent-skill, a new skill for assessing local-first dual-mode agentic system repositories
  • Evaluates 9 criteria across provenance/reproducibility, scientific capability, and user safety
  • Produces a structured Markdown report with pass/fail status and actionable recommendations

Test plan

  • Invoke the skill on a local agent repo and verify it produces a correctly formatted assessment table
  • Check that all 9 criteria (P1–P4, D1–D4, U2) are evaluated with file/line evidence

🤖 Generated with Claude Code

realmarcin and others added 2 commits February 17, 2026 19:59
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
… scope clarity

- Update description to specify Python-based local AI agent repositories
- Add clear triggers: "audit the repo", "check agent compliance", etc.
- Add Overview section defining "Local-First Dual-Mode Agentic System"
- Clarify target audience: computational biology/scientific workflows
- Add explicit "When NOT to use" guidance
- Expand Usage section with specific invocation examples

Addresses feedback:
1. Progressive disclosure: Description now includes WHAT (Python agents),
   WHEN (specific triggers), and WHO (scientific workflows)
2. Repository type: Explicitly states "Python-based" and provides context

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
@realmarcin

Copy link
Copy Markdown
Author

Closing to recreate PR from feature branch instead of main

@realmarcin realmarcin closed this Feb 19, 2026
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