openalice-voice-client
Architecture rating
Clean adoption of shared substrate (@openalicelabs/voice-client, valid manifest 100, 0 cycles, 0 dead imports, 0 naming violations) but drift 60 from 4 declared-unused TURN/VOICE env vars holds it below solid.
Strong structured docs (valid manifest, 14 @feature annotations, AGENTS.md + MIGRATION.md) but README is just an Atlas-card stub and 14 symbols miss docstrings — good metadata, weak in-code/prose narrative.
Only 2 test files (client.test.ts, types.test.ts) against 3 god-modules and a complexity-41 hotspot; no coverage figure and no evidence of reconnect/edge-case regression tests — thin for the lib's complexity.
WebRTC reconnect/teardown is the critical path but 10 broad catch blocks and max complexity 41 (avg 18.5) suggest fragile error handling; no visible timeout/retry/idempotency evidence and 4 stray console statements — graded down on unknowns.
Client-side WebRTC lib: no authn/authz surface, but TURN creds sit in NEXT_PUBLIC_* env (browser-exposed by design) and 10 broad/bare catch blocks swallow errors; no SSRF/injection/rate-limit/header signals — graded conservatively as mostly N/A for a library.
No persistence by design — manifest owns 0 tables, no migrations, no schema/FK/RLS; nothing to isolate or GDPR-scope, so axis is largely absent rather than poor.
No metrics/tracing/health probe (no https endpoint in manifest, health probe N/A for a lib); logging is 4 raw console.* calls with no structured logging — observability effectively absent.
source · auto-grader-2026-06-16
Telemetry ribbon · last 90 days
1 eventNo @feature annotations recorded in this repo yet.