QA Phase 5: Continuous QA Pipeline and Dashboardbacklog_item

qualityaifeature
1 min read · Edit on Pyrite

Parent

Subtask of qa-agent-workflows — Phase 5.

Problem

QA evaluations are ad hoc. Without a continuous pipeline, quality degrades over time as new entries skip review and old entries go stale. There is no visibility into overall KB health.

Solution

Automated pipeline that triggers Tier 1 on save (partially done), schedules batch Tier 2/3 runs, and surfaces results in a web dashboard.

Acceptance Criteria

  • ~~Post-save hook triggers Tier 1 validation automatically~~ Already done: `validate` param on `kb_create`/`kb_update` MCP tools + `qa_on_write: true` KB setting
  • Scheduled batch runs for Tier 2/3 (configurable frequency)
  • QA dashboard in web UI: verification rates, issue trends, coverage gaps
  • "Entries needing review" virtual collection with QA-based query
  • Integration with existing hooks system and collections
  • Dependencies

  • Phase 2 (qa-phase-2-qa-assessment-entry-type-and-storage) — assessment entries must exist
  • Hooks system (done)
  • Collections (done)
  • Capture lane validation
  • Files Likely Affected

  • Modified: `pyrite/services/qa_service.py` (batch scheduling)
  • New: `web/src/routes/qa/` (dashboard UI)
  • Modified: web frontend for QA views
  • Modified: `pyrite/server/endpoints/qa.py` (dashboard API)