Release 0.449.0

Version 0.449.0

Commits

• [a09576c] Fix convex-helpers pagination security advisory (#372)

Release 0.448.0

Version 0.448.0

Commits

• [62e8eb6] Curate native decision models and retain four LLM references (#370)

Release 0.447.0

Version 0.447.0

Commits

• [bf6ce73] Add Gemini 4 Argon to coding and decision evals (#369)

Release 0.446.0

Version 0.446.0

Commits

• [b8c7660] Add Muse Wonder and Muse Spark 1.3 (#368)

Release 0.445.0

Version 0.445.0

Commits

• [4a068da] Pass workflow inputs and model IDs to shell steps through env (#364)

Release 0.444.0

Version 0.444.0

Commits

• [e97ee60] Don't pass runner credentials to model-generated code (#363)

Release 0.443.0

Version 0.443.0

Commits

• [aa7b0ab] Compress guidelines from 7,010 to 4,146 tokens (#360)

Release 0.442.0

Version 0.442.0

Commits

• [68f5c0e] Add Claude Haiku 5.5 to the leaderboard and decision benchmark (#361)

Release 0.441.0

Version 0.441.0

Commits

• [4eda2e1] Remove the direct TypeSafe provider from decision evals (#358)

Release 0.440.0

Version 0.440.0

Commits

• [7a7ad98] Add GPT-6.1 Sol, and GPT-6 Luna to the decision benchmark (#353)