pr-tools

Use this agent when the user requests a deep, language-specific code review of a pull request, especially when they specify a programming language (Python, TypeScript, JavaScript, React, Rust) or pass a --language flag. This agent performs expert-level analysis using specialized devs skills to identify security vulnerabilities, logic errors, architectural issues, and code quality concerns. Examples: <example>Context: User wants to review a Python PR with expert analysis. user: "Review PR #45 in aaronbassett/api-service for Python code" assistant: "I'll use the pr-code-reviewer agent to perform a deep Python code review of that pull request." <commentary>The user explicitly requested a code review of a PR with a specific language (Python), which triggers this agent's specialized language-specific review capabilities.</commentary></example> <example>Context: Command invoked with language flag. user: "Can you review the code changes in PR #127 of myorg/backend --language rust" assistant: "I'll analyze that Rust PR with expert-level scrutiny." <commentary>The presence of the --language flag indicates the user wants language-specific analysis, which is exactly what this agent provides.</commentary></example> <example>Context: User wants comprehensive security and quality analysis. user: "I need a thorough review of PR #89 in frontend/dashboard focusing on TypeScript. Check for security issues, bugs, and architecture problems." assistant: "I'll use the pr-code-reviewer agent to conduct a comprehensive TypeScript code review covering security, bugs, and architectural concerns." <commentary>The request for thorough, multi-dimensional review (security, bugs, architecture) combined with a specific language perfectly matches this agent's capabilities.</commentary></example> <example>Context: Proactive suggestion during PR discussion. user: "This PR has some React components that need review before merging." assistant: "I can perform a deep expert review of those React components using language-specific best practices. Would you like me to analyze the PR?" <commentary>When discussing PR review needs with a specific language mentioned, the agent can proactively offer its specialized review capabilities.</commentary></example>

Use this agent when you need to consolidate multiple code review findings into a single PR comment, merge review results from different language reviewers, post or update GitHub PR comments with review summaries, or format comprehensive review reports. Examples: <example>Context: After running code-reviewer, todo-finder, and sdd-task-verifier agents, all outputs need to be combined into one PR comment. user: "Consolidate review findings and post to PR #123" assistant: "I'll consolidate all the review findings from the different agents and post them to the pull request." <commentary>The agent should trigger because we have multiple review outputs that need to be merged, deduplicated, formatted into a comprehensive report, and posted as a GitHub PR comment.</commentary> assistant: "I'll use the pr-comment-consolidator agent to merge the findings and update the PR comment."</example> <example>Context: Review agents have completed analysis. The orchestrator needs to create a final summary comment. user: "Update PR comment with these review results: {json_data}" assistant: "I'll process the review results and update the pull request comment." <commentary>This is an explicit request to consolidate and post review results. The agent should parse the JSON, organize findings by severity, and either update an existing Claude comment or create a new one.</commentary> assistant: "I'll use the pr-comment-consolidator agent to format and post the consolidated review."</example> <example>Context: A PR has been reviewed by multiple language-specific agents and todo-finder. user: "Create the final review report for PR #456" assistant: "I'll gather all the review findings and create a comprehensive report." <commentary>The agent should consolidate findings from all reviewers, deduplicate similar issues, organize by severity with appropriate indicators (critical, important, suggestions, highlights), and post via gh CLI.</commentary> assistant: "I'll use the pr-comment-consolidator agent to generate and post the final review report."</example> <example>Context: Review process completed with mixed results from different agents. user: "Post the consolidated review to the PR, including any failures" assistant: "I'll compile all results including any agent failures." <commentary>The agent needs to handle both successful and failed agent outputs, surface any analysis failures in the report, and ensure the PR author is aware of the complete review status.</commentary> assistant: "I'll use the pr-comment-consolidator agent to create a comprehensive report including failure notifications."</example>

Use this agent when you need to validate if a PR is reviewable, detect programming languages, or check if SDD (Spec-Driven Development) is in use. This agent performs fast initial triage of pull requests. Examples: <example>Context: User provides a PR number and repo for review. assistant: "I'll validate this PR first to check if it's reviewable and detect the languages used." <commentary>The agent should trigger to perform initial PR validation before any detailed review work.</commentary></example> <example>Context: User asks to review a PR. assistant: "Let me check if PR #45 in acme/api is ready for review." <commentary>Proactive triggering - before doing detailed review work, validate the PR status and gather metadata.</commentary></example> <example>user: "Validate PR #123 in user/repo" assistant: "I'll use the pr-validator agent to check if this PR is reviewable." <commentary>Explicit request to validate a PR triggers this agent directly.</commentary></example> <example>user: "Check if this PR uses SDD" assistant: "I'll analyze the PR to detect if Spec-Driven Development is in use." <commentary>Direct request for SDD detection should trigger this agent.</commentary></example>

Use this agent when you need to verify completed SDD (Spec-Driven Development) tasks in a pull request. This agent analyzes PR diffs to find newly checked tasks, parses task metadata, performs deep implementation reviews for P1 (MVP-critical) tasks, and validates file existence for other tasks. Examples: <example>Context: User is reviewing a PR that modifies specs/tasks.md with newly completed tasks. user: "Verify SDD tasks in PR #123" assistant: "I'll use the sdd-task-verifier agent to analyze the PR diff, extract completed tasks, and verify their implementation." <commentary>The user explicitly requests SDD task verification for a specific PR. This is the primary use case for the agent.</commentary></example> <example>Context: The pr-reviewer orchestrator has detected that specs/**/tasks.md was modified in a PR and needs task verification. user: "Review this PR" assistant: "I'll check the PR changes. I notice specs/features/tasks.md was modified with newly checked tasks. I'll use the sdd-task-verifier agent to verify these completed tasks." <commentary>The agent should proactively trigger when the orchestrator detects SDD task file modifications in a PR, even without explicit user request.</commentary></example> <example>Context: Developer wants to check if their P1 task implementations are complete before merging. user: "Check if my P1 tasks in this PR are properly implemented" assistant: "I'll use the sdd-task-verifier agent to perform deep verification of your P1 task implementations, checking the actual code against task descriptions." <commentary>User is concerned about P1 task quality, which requires the deep implementation review capability of this agent.</commentary></example> <example>Context: A PR comment thread discusses whether a task marked as complete actually has the implementation. user: "Is task T003 actually implemented in the changed files?" assistant: "I'll use the sdd-task-verifier agent to verify task T003's implementation status in this PR." <commentary>Specific task verification request - the agent can verify individual tasks by examining the PR diff and implementation files.</commentary></example>

Use this agent when you need to scan pull request changes for TODO, FIXME, HACK, XXX, NOTE, or BUG comments. This agent performs fast pattern matching on newly added lines to identify action items and technical debt markers. Examples: <example>Context: User wants to review a PR and check for TODOs. assistant: "I'll scan the PR for any TODO or FIXME comments in the newly added code." <commentary>The agent should trigger to find action items in PR changes before completing the review.</commentary></example> <example>Context: PR review is being orchestrated and needs TODO detection. assistant: "Let me find any TODO, FIXME, HACK, or other action items in PR #123." <commentary>Proactive triggering during automated PR review workflow to detect technical debt markers.</commentary></example> <example>user: "Find TODOs in PR #45" assistant: "I'll use the todo-finder agent to scan for TODO comments in the PR changes." <commentary>Explicit request to find TODOs triggers this agent directly.</commentary></example> <example>user: "Are there any FIXME comments in this PR?" assistant: "I'll scan the PR diff for FIXME and other action item comments." <commentary>Direct request for specific comment patterns should trigger this agent.</commentary></example>

This skill should be used when the user asks to "address the review", "discuss PR feedback", "plan fixes for the review", "create a plan to fix PR issues", or wants to brainstorm solutions for review findings. Use after a PR review has been posted to fetch the review comment and guide the user through creating an action plan using the brainstorming workflow.

This skill should be used when the user asks to "review this PR", "check the pull request", "review PR