Skip to content

Draft: Phase 1: evidence collector agent flow

Sofia Vistasrequested to merge
readiness-evidence-collector into main

What does this MR do?

Introduces the Readiness Evidence Collector - an AI workflow that automates checklist verification by scanning GitLab epics, issues, and architecture documents for evidence.

Why?

Manual readiness assessments require scanning numerous GitLab entities. This flow automates evidence collection and generates completion reports with citations.

What's included?

Flow: readiness_evidence_collector.yml (v1/ambient)

  • template_reader → reads checklist from structured folders
  • evidence_scanner → scans GitLab entities (epics + child issues)
  • evidence_analyzer → maps evidence to requirements
  • checklist_filler → generates markdown report

4 Prompts at v1.0.0 with optimized token budgets:

  • Scanner: 150K tokens (handles 30-50+ issues per epic)
  • Analyzer: 100K tokens
  • Filler: 80K tokens

Key Features

✅ Deep epic traversal (auto-discovers child issues)
✅ Multi-source evidence (descriptions, comments, docs)
✅ Status indicators (✓/⚠️/✗/❓)
✅ Evidence citations with hyperlinks

Testing

curl -X POST -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"workflow_definition": "readiness_evidence_collector/v1", ...}' \
  http://gdk.test:3000/api/v4/ai/duo_workflows/workflows

Merge request reports