moonshotai/kimi-k2-thinking#12

Merged

LearningCircuit merged 27 commits into

mainfrom

LearningCircuit-patch-1

Nov 9, 2025

Commits on Nov 9, 2025

moonshotai/kimi-k2-thinking
LearningCircuit
authored
Update ai-reviewer.sh
LearningCircuit
authored
feat: add comprehensive debug logging for thinking model responses
LearningCircuit
committed
fix: parse JSON from thinking model responses and add heart icon
LearningCircuit
committed
fix: simplify thinking model response handling

LearningCircuit
and
claude
committed
fix: escape forward slash in sed command
LearningCircuit
committed
debug: enable debug mode to see full AI response
LearningCircuit
committed
fix: remove markdown code blocks from AI response
LearningCircuit
committed
debug: remove content limits to see full AI response
LearningCircuit
committed
fix: resolve bash syntax error and enable markdown code block removal
LearningCircuit
committed
debug: output raw AI response to logs instead of file
LearningCircuit
committed
fix: properly handle thinking model responses with multiline regex
LearningCircuit
committed
fix: correct YAML indentation for Cleanup step
LearningCircuit
committed
fix: increase AI_MAX_TOKENS to 8000 to prevent response truncation
LearningCircuit
committed
fix: remove max_tokens limit to prevent response truncation
LearningCircuit
committed
refactor: address AI review recommendations
LearningCircuit
committed
style: remove emoji from header, keep heart in footer only
LearningCircuit
committed
style: use 'Friendly AI Reviewer' instead of 'F-AI-R' in footer
LearningCircuit
committed
fix: use constants consistently for review header and footer
LearningCircuit
committed
fix: increase AI_MAX_TOKENS to 64000 to prevent response truncation
LearningCircuit
committed
feat: include human comments and limit AI review context
LearningCircuit
committed
feat: include PR description and commit messages in review context
LearningCircuit
committed
docs: update README with accurate costs and new features
LearningCircuit
committed
fix: respect DEBUG_MODE flag for all debug output
LearningCircuit
committed
feat: improve AI review focus and conciseness
LearningCircuit
committed
refine: reduce repetition in AI reviews and clarify assumptions
LearningCircuit
committed
refactor: simplify review structure and clarify assumptions
LearningCircuit
committed