feat(agents): complete all 12 agents — OCR, Skill, Experience, Education, ATS, Judge, Gap, Interview, Bias, Fraud, Recommendation

feat(agents): add extraction agents #3-#5 (Skill, Experience, Education)

fix: resolve mypy errors in cv_parser agent

fix: add prometheus-client to runtime deps, make metrics import lazy

fix(ci): resolve ruff lint errors, mypy type issues

feat(providers): add HF Inference API adapter with T2 support

feat(agents): add CV Parser (#1), ATS Scoring (#6), retrieval funnel

feat(api): add screening run endpoints, wire agent registry

feat(db): add tenants, users, candidates, audit, chat, insights tables

feat: ONNX embedding, rate-limit scheduler, SSE streaming, reranker config

feat(security): enable Row-Level Security on all tenant-scoped tables

feat: add orchestrator, workflow runner, and service ports

feat: add /metrics endpoint, Supabase JWKS auth, Prometheus instrumentation

feat(agents): add Agent ABC, Semantic Matching judge (#7)

feat(db): add revised-stack tables — queue, cache, rate-limit, checkpoints, ATS, fraud, bias

Merge pull request #3 from Rzq12/refactor/stupid-simple-cleanups

refactor: simplify rate limiter to IP-only and add ChunkRepositoryProtocol

feat(models): add screening run and scoring tables

feat(api): expose the rubric template catalogue over HTTP

feat(api): add starter rubric templates per role family

feat(agents): add the JD Analyst that drafts a rubric from a job description

fix(models): declare the two pgvector/GIN indexes in ORM metadata

feat(agents): add LLM provider layer with failover and PII-tier gating

feat(scoring): implement the deterministic Phase 4 scoring core

fix(api): close eight defects found reviewing the rubric surface

feat(api): expose rubric authoring and approval endpoints

feat(infra): add rubric_versions and requirements migration

feat(rubric): add tenant-scoped rubric persistence

feat(rubric): add immutable rubric versioning with exact weight normalization

fix(search): connect ingestion to search and remove SQL injection

fix(db): declare content_tsv as TSVECTOR to match the migration

feat(search): add Phase 2 semantic search with pgvector and hybrid retrieval

fix(db): resolve ResumeChunk vector type and declare is_parent column

feat(security): strip and score prompt injection on every ingest

feat(db): add Alembic migration suite with initial schema

fix(security): enforce RBAC and close five denial-of-service vectors

feat: scaffold CV screener serving layer with ingestion pipeline, REST API routers, repositories, storage, parsing, security, and integration/unit test foundations

feat(api): add Phase 0-1 backend for identity and document ingestion