The PR Agent validates fixes through automated testing, explores alternatives using multiple AI models, and synthesizes everything into actionable recommendations.

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#1e1e2e', 'primaryTextColor': '#cdd6f4', 'primaryBorderColor': '#45475a', 'lineColor': '#6c7086', 'secondaryColor': '#313244', 'tertiaryColor': '#181825'}}}%%
flowchart LR
    subgraph gate ["🧪 GATE"]
        direction TB
        G1[Verify tests fail without fix]
        G2[Verify tests pass with fix]
        G1 --> G2
    end
    
    subgraph review ["🤖 PR REVIEW"]
        direction TB
        R1[Pre-Flight: gather context]
        R2[Try-Fix: 4 models explore alternatives]
        R3[Report: write recommendation]
        R1 --> R2 --> R3
    end
    
    subgraph post ["📊 POST"]
        direction TB
        P1[Post AI summary comment]
        P2[Apply agent labels]
        P1 --> P2
    end
    
    gate --> review --> post
    
    style gate fill:#1e1e2e,stroke:#89b4fa,stroke-width:2px,color:#cdd6f4
    style review fill:#1e1e2e,stroke:#cba6f7,stroke-width:2px,color:#cdd6f4
    style post fill:#1e1e2e,stroke:#a6e3a1,stroke-width:2px,color:#cdd6f4

Every fix is tested. The agent doesn't theorize—it implements each approach, runs tests, and reports what works.

For the best results, start in plan mode to create and review a detailed plan before execution:

1. Enter plan mode: Press Shift+Tab or use /plan
2. Request a review plan:

   /plan review PR #12345 - create a detailed plan for the review
   

3. Review the plan: Copilot will create a structured plan. Review the steps and make adjustments.
4. Exit plan mode: Press Shift+Tab to switch back to execution mode
5. Execute the plan:

   proceed with the plan
   

This workflow gives you control over the review approach before any actions are taken.

If you prefer to start immediately without planning:

# Start GitHub Copilot CLI
copilot

# Ask it to review a PR
please review PR #12345

The pipeline is orchestrated by Review-PR.ps1, which runs 4 steps:

Creates a clean review branch from main and squash-merges the PR onto it. If there are merge conflicts, the script posts a comment on the PR and exits.

Runs verify-tests-fail.ps1 to confirm tests actually catch the bug:

• Tests must fail without the fix (on the main baseline)
• Tests must pass with the fix applied

If the PR has no tests, the gate can invoke write-ui-tests or write-xaml-tests to create them. The gate result is passed forward as context.

Invokes the pr-review skill via Copilot CLI, which runs three phases:

Pre-Flight — Reads the linked issue, PR description, comments, and classifies changed files. No code analysis yet — just context gathering.

Try-Fix — The core of the review. Four AI models explore independent fix ideas sequentially, each learning from prior attempts:

Each model:

1. Generates an independent fix idea (without seeing the PR's solution first)
2. Implements it
3. Runs tests to verify

After all 4 attempts, a cross-pollination round asks each model if it has new ideas based on all results so far. This repeats until all models say "NO NEW IDEAS" (max 3 rounds).

The best fix is selected by comparing all passing candidates on simplicity, robustness, and codebase consistency.

Report — Writes the final review recommendation (APPROVE or REQUEST CHANGES) with a comparison of all fix candidates.

Posts the review as a collapsible comment on the PR, including the gate result, try-fix comparison table, and final recommendation.

Parses the phase output files and applies s/agent-* labels to the PR:

All phase output is written to CustomAgentLogsTmp/PRState/{PRNumber}/PRAgent/{phase}/content.md. This allows resuming interrupted sessions and tracking fix attempts.

• Agent Skills - All available skills
• Write Tests Agent - Test creation agent
• Learn From PR Agent - Post-PR learning
• Sandbox Agent - Manual testing agent