openalice-presets

Valid manifest with clear summary, AGENTS.md + README present, 10 @feature annotations (features 100/100) — only weaknesses are 4 missing docstrings and a thin roadmap (30/100); architecture intent well documented.

Single test file (src/__tests__/presets.test.ts) covering 10 features; no coverage metric or test count available and no CI signal ('?'), so critical-path coverage of scoring-weight/rate-budget logic is unverified — graded conservatively.

Pure library with no runtime process — no timeouts/retries/graceful-shutdown applicable; code is clean (avg complexity 1.0, 0 cycles, 0 bare excepts) but no idempotency or error-handling signals to evaluate; conservative pass.

Config-only npm package (@openalicelabs/presets) with no auth/network surface, no secrets, zero runtime deps — but no declared input-validation or schema-guard on preset payloads (system prompts/rate budgets), so injection-into-prompt safety is unverified; graded conservatively as low-surface-but-unproven.

Owns zero tables, no migrations, no DB/RLS/FK — data is static TS preset objects shipped as a package; no persistence layer to grade, so scored conservatively; no GDPR/PII concerns present.

No HTTPS endpoint (health probe fails: 'no https endpoint in manifest exposes'), no metrics/tracing/structured-logs — expected for a static preset package, but zero runtime observability signal so scored low.