bishalprasad321/PR Pilot Review

Intelligent multi-model consensus PR reviewer - Automated code reviews with consensus from multiple AI models.

Trust Signals

name	description	required	default
github_token	GitHub token with permissions to read PR details and post reviews.	yes	—
llm_provider	Language model provider to use for PR review. Supported values: gemini or groq. Defaults to groq.	no	groq
llm_api_key	API key for the chosen LLM provider. Use a provider-specific API key such as GROQ_API_KEY for groq.	no	—
llm_provider_url	Optional base URL for the chosen LLM provider. For Groq, use https://api.groq.com/openai/v1 (uses /chat/completions endpoint).	no	—
reviewer_models	Comma-separated list of 3 review models. Use provider-specific model IDs such as llama-3.1-8b-instant or openai/gpt-oss-20b.	no	llama-3.1-8b-instant,openai/gpt-oss-20b,llama-3.3-70b-versatile
judge_model	Model used for consensus judgment when reviewer opinions differ. Use provider-specific model IDs such as openai/gpt-oss-120b.	no	openai/gpt-oss-120b
max_consensus_rounds	Maximum number of consensus rounds before forcing judge decision. Each round queries all 3 reviewers again.	no	3
inline_comments_enabled	Enable inline code comments on specific lines where issues are found. If false, only summary comment posted.	no	true
max_diff_lines	Maximum diff lines to process. If diff exceeds this, will process in chunks or summarize.	no	5000
enable_incremental_diff_processing	Enable incremental diff processing to handle large diffs by processing them in chunks.	no	true
debug	Enable debug mode for verbose logging. Shows consensus rounds, model interactions, etc.	no	false

name	description
review_decision	Final consensus decision from judge model (APPROVE, REQUEST_CHANGES, or COMMENT)
consensus_reasoning	Judge model's reasoning for the consensus decision
consensus_round	Which round (1-3) consensus was achieved, or 3 if forced after max rounds
review_id	GitHub PR review ID if successfully submitted