Code Quality reference
Analyze code for style violations, correctness issues, potential bugs, and architectural problems using specialized review agents.
Classes
| Class |
Description |
CodeReviewWorkflow |
SDK-native code review with four specialized subagents |
CodeReviewWorkflow
| Method |
Parameters |
Returns |
Description |
__init__ |
**kwargs: Any |
None |
Initialize the code review workflow |
execute |
**kwargs: Any |
WorkflowResult |
Run the multi-agent code review process |
Constants
| Constant |
Values |
Description |
_SUBAGENT_NAMES |
{'security-reviewer', 'quality-reviewer', 'perf-reviewer', 'architect-reviewer'} |
Names of the four specialized review agents |
_SYSTEM_PROMPT |
'You are a senior code review orchestrator. You coordinate four specialized subagents to produce a unified code review report. Be thorough but concise. Cite file paths and line numbers when possible.' |
System prompt for the orchestrating agent |
_TASK_PROMPT_TEMPLATE |
'Review the codebase at {path} using the four specialized subagents below. Each subagent should focus on its domain and report findings as structured markdown.\n\nAfter all subagents finish, synthesize their findings into a single report with these sections:\n\n## Summary\nOverall code health score (0-100) and a 2-3 sentence executive summary.\n\n## Security\nFindings from the security reviewer.\n\n## Quality\nFindings from the quality reviewer.\n\n## Performance\nFindings from the performance reviewer.\n\n## Architecture\nFindings from the architecture reviewer.\n\n## Suggestions\nActionable next steps ordered by priority.' |
Template for structuring the multi-agent review task |
Source files
src/attune/workflows/code_review.pysrc/attune/workflows/code_review_*.py
Tags
review, quality, bugs, lint