# Changelog

## Score

- CAI 54 → 54 (-0.6)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.4) — scores are not directly comparable.

## Lenses

- Code Health 89 → 85 (-4.1)
- Architecture 88 → 95 (+7.8)
- Maturity 78 → 79 (+1.5)
- Readiness 66 → 66 (+0.1)
- Security 32 → 31 (-1.9)

## Resolved (18)

- Duplicated block (13 lines × 3) (lib/req_llm/providers/minimax/images_api.ex)
- Duplicated block (8 lines × 2) (lib/req_llm/provider/defaults.ex)
- Duplicated block (8 lines × 4) (lib/req_llm/provider/defaults.ex)
- Duplicated block (9 lines × 2) (lib/req_llm/provider/defaults.ex)
- Members sharing a duplicated core (4 members, 50+ identical tokens) (lib/req_llm/provider/defaults.ex)
- Members sharing a duplicated core (4 members, 50+ identical tokens) (lib/req_llm/provider/defaults.ex)
- Members sharing a duplicated core (6 members, 50+ identical tokens) (lib/req_llm/providers/minimax.ex)
- Off-boarding risk: anonymized user #1
- Orphaned file (lib/req_llm/providers/google/response_builder.ex)
- Outdated: igniter
- Outdated: jsv
- Outdated: phoenix_playground
- Outdated: splode
- Projects may be oversized for their cohesion
- Repeated repair: lib/req_llm/providers/anthropic.ex (lib/req_llm/providers/anthropic.ex)
- Repeated repair: lib/req_llm/providers/openai_codex.ex (lib/req_llm/providers/openai_codex.ex)
- Repeated repair: lib/req_llm/providers/zai_coding_plan.ex (lib/req_llm/providers/zai_coding_plan.ex)
- Usage.handle (cognitive 18) (lib/req_llm/step/usage.ex)

## New (203)

- Duplicated block (11–17 lines × 2) (lib/req_llm/providers/cohere.ex)
- Duplicated block (13 lines × 3) (lib/req_llm/providers/minimax/images_api.ex)
- Duplicated block (7 lines × 2) (lib/req_llm/evaluation/protocol.ex)
- Duplicated block (8 lines × 2) (lib/req_llm/providers/catalog_gateway.ex)
- Duplicated block (8 lines × 2) (lib/req_llm/providers/google_vertex.ex)
- Duplicated block (8 lines × 3) (lib/req_llm/providers/catalog_gateway.ex)
- Duplicated block (8 lines × 4) (lib/req_llm/provider/defaults.ex)
- Duplicated block (9 lines × 2) (lib/req_llm/provider/defaults.ex)
- Edited copies of one member (3 members, 12+ corresponding lines) (lib/req_llm/providers/anthropic.ex)
- Edited copies of one member (3 members, 14+ corresponding lines) (lib/req_llm/providers/anthropic/response.ex)
- Edited copies of one member (3 members, 2+ corresponding lines) (lib/req_llm/provider/defaults.ex)
- Edited copies of one member (3 members, 25+ corresponding lines) (examples/scripts/json_schema_examples.exs)
- Edited copies of one member (3 members, 3+ corresponding lines) (lib/req_llm/providers/anthropic.ex)
- Edited copies of one member (3 members, 30+ corresponding lines) (lib/req_llm/providers/amazon_bedrock/anthropic.ex)
- Edited copies of one member (3 members, 36+ corresponding lines) (lib/req_llm/providers/fireworks_ai.ex)
- Edited copies of one member (3 members, 4+ corresponding lines) (lib/req_llm/auth.ex)
- Edited copies of one member (3 members, 4+ corresponding lines) (lib/req_llm/streaming.ex)
- Edited copies of one member (3 members, 4+ corresponding lines) (tutorial/agents/06_memory_context.exs)
- Edited copies of one member (3 members, 8+ corresponding lines) (examples/lib/req_llm/examples/helpers.ex)
- Edited copies of one member (3 members, 9+ corresponding lines) (lib/req_llm/ocr.ex)
- …and 183 more

## Changes since last survey

- 30 commits — 17 feature/other, 13 fixes

## By area

- lib/req_llm — 13 commits
- (root) — 9 commits
- test/req_llm — 5 commits
- test/support — 2 commits
- guides/anthropic.md — 1 commit

## Notable commits

- fix: fix(anthropic): preserve cache counter presence (#1230)
- fix: fix(deps): require patched JSV and unblock offline CI (#1235)
- fix: fix(google): carry raw finish reason onto streamed responses (#1266)
- fix: fix(google): preserve non-streaming function-call thought signatures (#1246)
- fix: fix(google): send responseJsonSchema to Gemini 4 models (#1269)
- fix: fix(google): update JSON schema fixture bodies (#1270)
- fix: fix(google_vertex): handle non-2xx responses (#1237)
- fix: fix(mc): repair Groq test URL and mix mc replay routing (#1251)
- fix: fix(mistral): accept low and medium reasoning effort (#1228)
- fix: fix(streaming): normalize context window errors (#1262)
- fix: fix(timeout): stop retries after the total deadline (#1244)
- fix: fix: derive streamed tool-call argument flags from the assembled fragments (#1256)
- fix: fix: keep the invalid_arguments flag on buffered tool calls (#1254)
- change: build(deps): require ex_aws_auth ~> 1.6 (#1264)
- change: chore(deps): update Zoi to 0.19 (#1240)
- change: chore: release version v1.30.0
- change: deps(deps): bump jsv from 0.26.2 to 0.27.0 (#1259)
- change: deps(deps-dev): bump websock_adapter from 0.6.0 to 0.7.0 (#1239)
- change: deps(deps-dev): bump websock_adapter from 0.7.0 to 0.7.1 (#1258)
- change: deps: refresh dependency lock and security patches
- …and 10 more
