Interrogate

Run one keel:critic reviewer per configured model to adversarially review code changes, plus the codex lane when it is configured. The adversarial signal comes from model diversity, not assigned personas. The deliverable is a synthesized verdict. Do NOT auto-apply changes.

Step 1, Scope

Identify what to review from context:

• If the user points at specific files or a diff, use that.
• If on a feature branch, run git diff main...HEAD (or the appropriate base branch) for the full changeset.
• If the user's message references recent work, gather the relevant files.

Package the diff or file contents with any surrounding context files the reviewers need to understand the code.

Step 2, Intent

Before spawning reviewers, state the intent in one clear paragraph from the user's message, commit messages, the PR description if one exists, and the code. If you're unsure about the intent, ask the user before proceeding.

Step 3, Spawn Reviewers

Launch all reviewers at once, one per entry on the interrogate reviewers line, labeled Reviewer A, B, and so on to match the entry count. Claude reviewers are Agent calls in a single message. The codex reviewer is a Bash command started in the same message. Read ${CLAUDE_CONFIG_DIR:-$HOME/.claude}/keel-models.md if it exists; use the role's line, else the table defaults.

Each Claude reviewer gets subagent_type: "keel:critic" and model set to its configured entry, or its table default when there is no configured line. Omit model for an inherit, inherit-parent, or auto entry, so that reviewer runs on the parent model. Its prompt opens with "Read-only: do not edit files. Lens: correctness, through the interrogate rubric and code-quality lens below. Use the brief's severity scale."

If the Agent tool rejects a configured entry, run that reviewer on Reviewer A's default and say so. If it rejects a table default, omit model for that reviewer, say so, and open a separate PR to update the default table. Do not block the review on the model issue. Never treat an alias entry as a rejected model or apply either fallback to it.

The codex entry is a cross-vendor lane, not a subagent. Run command -v codex first. If it fails, skip the lane and name it as skipped in the Reviewers list. Otherwise write the filled template below to a file and run Codex read-only from the repository root, in the background so it runs alongside the subagents:

codex exec --sandbox read-only --ephemeral --cd "$(git rev-parse --show-toplevel)" \
  --output-last-message "$tmp/codex-review.md" - < "$tmp/reviewer-brief.md"

$tmp is a scratch directory you create for the run. The brief goes in on stdin because codex exec review --base <branch> refuses a custom prompt, and every reviewer gets the same brief. When the command finishes, read $tmp/codex-review.md and treat it like any other reviewer's output.

Read references/reviewer-prompt.md and fill in the same template for every reviewer with the stated intent, the diff or file contents, the rubric from references/rubric.md, and the code-quality lens from references/code-quality-review.md. Pick each subagent's effort for the job, per the Subagents section of the lead skill.

Step 4, Synthesize

As results come back, build a unified picture. Parse every reviewer's findings. Merge findings that describe the same issue differently, and note which models raised each one. Findings two or more models raised independently are the highest signal. Read a lone model's finding, but weight it accordingly. Note disagreements. If one model flags something and another explicitly says the opposite, that is context for the verdict.

Step 5, Lead Judgment

You are the lead reviewer, a pragmatic senior engineer, not a neutral aggregator. Read references/lead-judgment.md for the full framework. Put every finding in one of four buckets, with the model(s) that raised it and a one-line rationale.

• Act on. Real issues affecting correctness, security, or maintainability given the actual goals. They would block a real PR.
• Consider. Legitimate, but you're not sure they outweigh the cost of addressing them now. Worth the user's attention.
• Noted. Technically valid but not actionable. Context-dependent, premature optimization, or low-impact at this stage.
• Dismissed. Wrong, nitpicky, or missing context, with a brief reason.

Output Format

Intent

[The stated intent paragraph from Step 2]

Reviewers

• Reviewer [label]: [model name], [N findings] (one bullet per reviewer)

Act On

[Each: description, which models raised it, why it matters.]

Consider

[Each: description, which models raised it, the tradeoff.]

Noted

[Brief list.]

Dismissed

[Each with a brief rationale.]

Agreement Map

[Where models agreed and diverged, and what that pattern tells us.]