A platform team is introducing Copilot cloud agent into an existing SDLC. They want to avoid common anti-patterns such as vague tasks, unreviewed draft pull requests, validation gaps, and agent changes that bypass repository ownership controls. Which four actions should you perform in order? 1. Assign the issue to Copilot with repository-specific context 2. Merge after Copilot reports the task is finished 3. Define the issue task, scope, expected output, and acceptance checks 4. Review the generated draft pull request and diff 5. Add or verify required checks and owner review controls 6. Disable path filters so every workflow always runs

The correct order addresses the anti-patterns by establishing clear requirements and controls before agent involvement. First, "Define the issue task, scope, expected output, and acceptance checks" (3) prevents vague tasks and ensures the agent understands the goal. Next, "Add or verify required checks and owner review controls" (5) establishes necessary validation and prevents the agent from bypassing ownership controls. Then, "Assign the issue to Copilot with repository-specific context" (1) provides the agent with the information it needs to begin work effectively. Finally, "Review the generated draft pull request and diff" (4) addresses unreviewed draft pull requests and allows for human oversight before merging. The other options either misorder these crucial steps, leading to the anti-patterns, or include irrelevant actions like disabling path filters.