review

name	review
description	All quality reviews for manuscripts, methods, code, talks, and replication packages. Use when the user asks for code review, paper review, causal audit, proofreading, peer review simulation, stress testing, or a weighted quality score.

Review

Unified review workflow that routes to the appropriate critic or referee roles based on the target and requested mode.

Read First

• AGENTS.md
• plugins/clo-author/references/runtime-activation.md
• plugins/clo-author/references/codex-workflow.md
• plugins/clo-author/references/critic-runtime-contract.md
• plugins/clo-author/references/content-invariants.md
• plugins/clo-author/references/quality-scoring.md
• plugins/clo-author/references/logging.md
• plugins/clo-author/references/agent-catalog.md
• plugins/clo-author/references/domain-profile.md
• plugins/clo-author/references/journal-profiles.md when journal calibration is requested

Routing Logic

Auto-detect by target

• manuscript source -> comprehensive manuscript review
• R, Python, Stata, or Julia script -> code review
• talk source -> storyteller-critic audit

Explicit modes

• code review
• data-engineering review
• manuscript review
• methods-only review
• proofread / polish
• peer review simulation
• hostile stress test
• cross-language replication review
• full weighted review

Mode Details

Comprehensive Manuscript Review

Dispatch in parallel when appropriate:

1. strategist-critic for the causal design audit
2. writer-critic for manuscript polish
3. Verifier for mechanical checks

Compute the weighted outcome using plugins/clo-author/references/quality-scoring.md.

Code Review

1. Run plugins/clo-author/hooks/lint-scripts.sh [file|dir] first.
2. Dispatch plugins/clo-author/references/active-critics/coder-critic.md in read-only mode.
3. Save the report and quality outcome before proposing revisions.

Data-Engineering Review

1. Run plugins/clo-author/hooks/lint-scripts.sh [file|dir] first when scripts exist.
2. Dispatch plugins/clo-author/references/active-critics/data-engineer-critic.md in read-only mode.
3. Save the report and quality outcome before proposing revisions.

Methods-Only Review

Dispatch plugins/clo-author/references/active-critics/strategist-critic.md standalone and run the four-phase econometric review.

Proofread / Polish

Dispatch plugins/clo-author/references/active-critics/writer-critic.md standalone, with content-invariants.md as the hard invariant layer for mechanical issues.

Full Peer Review

The peer-review route is explicit: editor desk review -> referee dispatch -> editorial decision.

1. Editor runs a desk review and assigns referees.
2. domain-referee and methods-referee review independently and blindly.
3. Editor synthesizes the reports into FATAL / ADDRESSABLE / TASTE and produces the decision memo.

Save outputs to quality_reports/ with separate artefacts for desk review, referee reports, and the editorial decision.

Hostile Stress Test

Use the same peer-review routing but instruct the editor to select more adversarial calibration when the user asks for a stress test.

Cross-Language Replication Review

If the user asks for replication review:

• dispatch a second implementation in the requested language
• review both implementations
• compare output parity against the active tolerances before issuing a verdict

Reporting Rules

• For manuscripts and talks, cite invariant numbers when the issue is mechanical and non-negotiable.
• When reporting a weighted score, ground it in the active scoring rubric.
• Save findings before proposing revisions.
• Append a research-journal entry whenever the review changes project status or produces a formal gate outcome.

Outputs

• Review memo in quality_reports/
• Weighted score or pass/fail result when applicable
• Peer-review decision memo when the editor/referee flow runs
• Clear next-step recommendation grounded in the saved review artefacts

Do not edit source files while running the review workflow. Review first, then route revisions explicitly.