tree: 1fff9b701a27330b759760d9552e86e982032b0f
  1. step-5-security-checklist/
  2. README.md
tools/skill-evals/evals/write-skill/README.md

write-skill evals

Behavioral evals for the write-skill skill.

Suites (5 cases total)

SuiteStepCasesWhat it covers
step-5-security-checklistStep 5 (apply security checklist)5reads-public-pr-comments, reads-gmail-private, local-only (no external reads), prompt-injection-in-description, reads-public-issue-bodies

Run

# All cases
uv run --project tools/skill-evals skill-eval \
    tools/skill-evals/evals/write-skill/

# Single suite
uv run --project tools/skill-evals skill-eval \
    tools/skill-evals/evals/write-skill/step-5-security-checklist/fixtures/

# Single case
uv run --project tools/skill-evals skill-eval \
    tools/skill-evals/evals/write-skill/step-5-security-checklist/fixtures/case-1-reads-pr-comments