feat(agent): forward model settings and recover from thinking stalls [PC-4672] - #1017

Merged

tudormatei1 merged 11 commits into

mainfrom

feat/model-specific-settings

Sep 2, 2026

Commits on Sep 2, 2026

feat(agent): forward model settings and skip forced tool_choice under Claude thinking
tudormatei1
committed
b30c98d
feat(agent): retry a stalled thinking turn with a forced extraction call
tudormatei1
committed
ba04cd9
feat(agent): tolerate one tool-less turn before failing the run
tudormatei1
committed
a33cc58
refactor(chat): centralize thinking config handling in chat/thinking.py
tudormatei1
committed
0ddd334
refactor(agent): fold licensing error mapping into exceptions/llm
tudormatei1
committed
bc770c5
fix(agent): keep provider error bodies out of run telemetry
tudormatei1
committed
0b574d8
refactor(agent): apply review feedback on thinking and extraction
tudormatei1
committed
c832e27
fix(agent): update analyze_files_tool import after the licensing merge
tudormatei1
committed
502f437
revert(agent): restore provider http error mapping to main behaviour
tudormatei1
committed
2675455
chore: bump to 0.16.15 and require uipath-langchain-client>=1.18.3
tudormatei1
committed
56649a9
test: type-annotate the forced-extraction and llm-node stubs
tudormatei1
committed
6828913