Peer Review

Independent peer review via codex. Runs /codex-exec in read-only mode with a review prompt that was passed in or derived from context.

Step 1: Identify the Review Prompt

Determine the review prompt:

• If a review prompt was passed in (in the Agent task context), use it as-is
• If no prompt was provided (standalone invocation), determine what to review from conversation context and build a prompt using /codex-exec XML tags (<task>, <structured_output_contract>, and optionally <dig_deeper_nudge>)

When building a standalone prompt, use P0-P3 priorities: P0 (fundamentally flawed), P1 (significant gap), P2 (moderate issue), P3 (minor improvement). Example:

<task>
Review the following for issues that would cause incorrect behavior or get an implementer stuck.
</task>

<structured_output_contract>
For each issue: (1) problem, (2) location, (3) impact, (4) suggested fix, (5) priority P0-P3.
If no issues, state that it looks sound.
</structured_output_contract>

Step 2: Run /codex-exec Skill

Run the /codex-exec skill in read-only mode with the review prompt.

When the prompt covers multiple review dimensions (e.g., correctness, security, quality, API usage), wrap the task list with parallel fan-out instructions so codex delegates each dimension to a separate sub-agent. Add to the prompt:

Delegate each review dimension to a separate sub-agent using spawn_agent so they run in parallel. Wait for all agents to complete, then synthesize their findings into a unified report.

See /codex-exec references/parallel-execution.md for details on codex parallel execution.

Step 3: Output Findings

Output the codex findings.