calibrate(jury): 4A characterizes v1.1.1 residual as model-class-specific

calibrate(jury): v1.1+v1.1.1 — fix weighting bugs; recency-position paraphrase clause

rubric: clarify groundedness reference scope (snippets-only) for v1.1 gold (#20)

docs+build: judge-layer v1 coupled-artifact updates

docs(decisions): promote cold-start falsified-assumption and audit-path incident entries, add three-regimes latency refinement

docs(decisions): add entry on named residual risks and scope limits verdict discipline

docs+test: round-2 incident response — Google API key format scrub

docs: incident response + SHA remap after credential-exposure history rewrite

docs: defer HF Space rename — outstanding applications reference current URL

docs: step 8.1 — tagline reframe + README honest-scope + rename closure

feat(eval): K8s refusal_threshold sweep against 25Q set — 0.015 validated

docs: step 5 follow-up — parallel-tracks list + post-authoring observations

feat(eval): Week 1 step 5 — 25-question K8s golden dataset + grounded_refusal fix

docs: Phase 1 gate closure + stale-wording corrections (cross-cutting #3)

docs(eval): Fix 2 SearchTool query expansion — attempted and reverted

docs(eval): Fix 1 counterfactual prompt clause — attempted and reverted

chore(eval): pin gpt-4o-mini snapshot + wire fastapi golden_dataset + pre-commit tolerances

feat(eval): K8s refusal_threshold 0.02 → 0.015 empirical calibration

feat(eval): K8s first pilot run + flavor-B empirical decisions

docs: decisions for multi-corpus refactor

docs: add decisions for monitor mode, SSE events, vanilla JS

docs: fix test count in Testing section, add auth decision, reorder entries

docs: add security architecture section to README and DECISIONS.md

feat: infrastructure sprint — vLLM/Modal, Helm, Terraform (#8)

fix: remove stale V1 docs, update DECISIONS.md for V2

feat: implement Anthropic Claude provider

feat: add SQLite conversation sessions with session_id

feat: add provider retry with backoff and API rate limiting

feat: add cross-encoder reranking with feature flag

feat: add grounded refusal gate based on retrieval score threshold

feat: Day 8 — README with architecture, API docs, eval guide + DECISIONS.md