← ClaudeAtlas

aria-skill-testlisted

Regression harness for the aria-skills repo. Runs every shipped skill in dry-run mode + asserts clean output (no traceback, no credential leaks, valid SKILL.md frontmatter, README.md present). Text report by default; --junit-xml flag for CI integration. Use when the user says "test the skills", "regression check", "before push", "run the test harness", or any time before opening a PR or pushing to main.
iansteitz1-eng/aria-skills · ★ 0 · Testing & QA · score 60
Install: claude install-skill iansteitz1-eng/aria-skills
# aria-skill-test Regression harness. Pythonic + zero-dependency. Tests structural correctness across every skill in the repo without making any real API calls. ## When to use - Before pushing changes to the public repo - In CI on every PR (use `--junit-xml` flag) - After running `aria-skill-template` to confirm the generated stub passes - After editing any skill script — quick sanity check ## What gets checked Per skill directory: - SKILL.md frontmatter parses (YAML between `---` markers) - `name` field in frontmatter matches directory name - `description` field is 40-600 chars - README.md exists - Executable script found (snake_case or kebab-case .py / .sh) - Dry-run exits 0 (clean) or 2 (needs creds with clear FATAL message) - No Python traceback in stderr - No credential leak patterns in output (Anthropic / Stripe / Resend / Twilio / Cloudflare / EL / AWS / GitHub PAT) ## CLI ```sh python3 aria_skill_test.py # all skills python3 aria_skill_test.py --skill <name> # one skill python3 aria_skill_test.py --skip <name> --skip <name> # exclude python3 aria_skill_test.py --junit-xml results.xml # CI mode python3 aria_skill_test.py --repo-root /path/to/repo # alternate ``` ## See also - **[BUILDER_GUIDE.md](../BUILDER_GUIDE.md)** — the patterns enforced - **[aria-skill-template](../aria-skill-template/)** — generator paired with this harness ## License Apache 2.0