A team uses Copilot cloud agent to investigate a flaky checkout failure. The agent posts several intermediate observations in chat, changes two test files, and opens a draft pull request. A reviewer joins the next day and cannot see why one failing test was treated as unrelated. The team wants future agent sessions to preserve decisions where reviewers can audit them. Which option best mitigates the context-drift risk?

Recording decisions directly in the issue and the draft pull request (PR) provides a persistent, auditable record of the agent's rationale. This allows reviewers to understand why specific actions were taken, like treating a failing test as unrelated, even if they join the process later. This mitigates context drift by embedding the decision-making process within the development artifacts themselves. Storing reasoning only in Copilot Chat history is ephemeral and not easily accessible for future review. Adding a status badge to the README is for high-level status, not detailed decision logging. Re-running the same prompt before every review is inefficient and doesn't guarantee the agent will produce the exact same reasoning or actions, nor does it provide a persistent audit trail.