Patch Changes
-
#780
bf7b08bThanks @caseyg! - feat: add ibm bob as a provider and coding-agent host -
#869
9b35620Thanks @radekfojtik! - fix: add claude opus 5 to the anthropic and vertex model lists -
#870
f71581bThanks @pranaypolishetti26! - feat: add kiro coding-agent integration -
#801
2c90e13Thanks @ousamabenyounes! - feat: map gemini reasoning effort to thinking level -
#777
cf0700fThanks @danielsogl! - fix: import AGENTS.md into the managed CLAUDE.md block -
#863
fc62fb6Thanks @HwangJohn! - fix: require Node.js 22.22.0 or newer -
#788
4cd2e5fThanks @HwangJohn! - feat: let openai-compatible opt into reasoning effort -
#861
1692ebbThanks @HwangJohn! - Normalize redundant nested text content arrays for OpenAI-compatible Chat Completions requests so strict vLLM servers accept replayed tool output. -
#862
3994b9aThanks @HwangJohn! - fix: coerce roleless repository worker stream messages -
#864
2c6367eThanks @HwangJohn! - fix: retry transient openrouter provider 404 errors -
#848
054db7cThanks @kowshikdev! - fix: only restamp page-manifest entries a run actually regenerated -
#872
11cc526Thanks @tuandinh0801! - fix: instruct repository planner to invoke submit_plan directly without conversational text -
#856
590b1bbThanks @radekfojtik! - fix: route Vertex AI xAI Grok model IDs to the OpenAI-compatible surface -
#859
1b84ea8Thanks @HwangJohn! - fix: ignore windows ctime drift while fingerprinting -
#847
f4ca24fThanks @kowshikdev! - fix: add OPENAI_COMPATIBLE_STREAM_MESSAGES_ENV_KEY to MANAGED_ENV_KEYS -
#881
8175e29Thanks @marliechorgan! - fix: report a claim whose evidence now traverses a symbolic link as unresolved instead of aborting the update -
#885
7f81688Thanks @green-creeper! - fix: tighter redaction of API keys in CredentialDiagnosticsPanel