```items id migration-workbench title Migration Workbench group infra "Layer · Infrastructure" group parsing "Parsing" item T-101 "Define parsed policy schema" estimate: 2d item T-102 "Apigee XML bundle parser" estimate: 3d depends: T-101 item T-103 "Tibco BW project folder parser" estimate: 3d depends: T-101 item T-104 "Parser integration tests (sample bundles)" estimate: 2d depends: T-102 T-103 group database "Database" item T-201 "Postgres + pgvector schema setup" estimate: 1d depends: T-101 group ai "Layer · AI Pipeline" group rag "RAG" group apigee-corpus "Apigee Corpus" item T-202A "Apigee policy doc ingestion" estimate: 1d depends: T-201 item T-202B "Apigee migration guide ingestion" estimate: 1d depends: T-201 group mule-corpus "Mulesoft Corpus" item T-202C "Mulesoft connector doc ingestion" estimate: 1d depends: T-201 item T-202D "Mulesoft migration guide ingestion" estimate: 1d depends: T-201 item T-203 "Embedding pipeline (OpenAI text-embedding-3-small)" estimate: 2d depends: T-201 item T-204 "RAG retrieval endpoint + ranking" estimate: 2d depends: T-202A T-202B T-202C T-202D T-203 item T-205 "RAG retrieval smoke tests" estimate: 1d depends: T-204 group mapping "Mapping + Routing" item T-301 "Static type-based routing table" estimate: 1d depends: T-205 item T-302 "LLM mapping agent (pydantic-ai, OpenRouter)" estimate: 3d depends: T-301 item T-303 "Confidence scoring function" estimate: 2d depends: T-302 item T-304 "Mapping layer unit tests" estimate: 2d depends: T-303 item T-B101 "Empirical routing threshold calibration" estimate: 2d depends: T-303 item T-B102 "Routing config externalised (not hardcoded)" estimate: 1d depends: T-B101 group execution "Execution" item T-401 "Mulesoft project assembler (XML + DataWeave)" estimate: 4d depends: T-304 item T-402 "Aggregate confidence score + review flag logic" estimate: 1d depends: T-401 item T-403 "Output format decision (Studio project vs. zip)" estimate: 1d depends: T-401 item T-404 "Contract testing scaffold (pytest + pydantic)" estimate: 2d depends: T-401 group ux "Layer · Human Loop" group review-gate "Review Gate" item T-501 "Review queue data model (done / needs-review)" estimate: 1d depends: T-402 item T-502 "Review UX (CLI first pass)" estimate: 3d depends: T-501 item T-503 "Approve / correct / reject actions" estimate: 2d depends: T-502 group observability "Observability" item T-B201 "Migration state dashboard (per-API status)" estimate: 2d depends: T-501 group corpus-feedback "Corpus Feedback" item T-B301 "Write corrections back to pgvector index" estimate: 2d depends: T-503 item T-B302 "Supplemental quirks knowledge base" estimate: 1d depends: T-B301 ```