counsels-review

/counsels-review [--personas ]

Review an existing decision, policy, practice, or draft memo through multiple counsel lenses.

⚠️ NOT LEGAL ADVICE. This skill produces craft-register writing in the voice of canonical figures. The review is a reasoning exercise, not a representation by counsel. Real legal questions require real lawyers.

What this does

Dispatches counsel personas to review the artifact at <path>. Default panel runs three personas in parallel (RBG, Lessig, Rawls), each with a different lens. Output is a consolidated review — per-persona verdicts marked up, then a single recommended highest-leverage change.

When to use

• A draft memo needs scrutiny before being committed.
• An existing decision or policy benefits from a multi-lens review.
• An ethical question is "working but feels wrong" and needs named diagnoses.
• A practice has accumulated and needs an audit before further investment.

Not for: live legal review of a real matter (the disclaimer is not optional); pixel-level policy nits; parallel-debate format (filed for v1.0 as /counsels-debate).

Instructions for Claude

When this skill is invoked with a <path> argument and optional --personas:

1. Resolve the project root the same way /counsels-project-init does. Verify CLAUDE.md exists or warn that the counsel bible is missing.

2. Read the project specification:

   • CLAUDE.md, README.md, prior memos at counsel/memos/, prior reviews at counsel/reviews/, briefs at counsel/briefs/
   • The actual decision, policy, or practice the target concerns (if not already in the target)
   • .great-authors/project.md if cross-craft project

3. Read the target. <path> may be:

   • A single file (a memo, a policy, a draft brief)
   • A directory (a body of memos, a policy library)
   • A glob (counsel/memos/*.md)

   Read the target file(s) in full. Read adjacent context — sibling files, the underlying decision the memo addresses, prior reviews that motivated the current draft. The reviewer can't review what they can't see.

4. Resolve the panel. If --personas is given (comma-separated), use those. Otherwise default panel:

   • ruth-bader-ginsburg-counsel — constitutional / civil-rights / equal-protection lens
   • lawrence-lessig-counsel — regulability / digital-context / four-modalities lens
   • john-rawls-counsel — justice-as-fairness / veil-of-ignorance lens

   Common alternative panels:

   • Constitutional debate: ruth-bader-ginsburg-counsel + antonin-scalia-counsel — progressive vs. originalist (the canonical disagreement)
   • Privacy / antitrust review: louis-brandeis-counsel + tim-wu-counsel — the right to be let alone meets the curse of bigness
   • Regulatory review: cass-sunstein-counsel + lawrence-lessig-counsel — choice architecture meets code-as-law
   • Ethics / political philosophy review: hannah-arendt-counsel + john-rawls-counsel — the public realm meets the veil
   • Civil rights litigation strategy review: thurgood-marshall-counsel + ruth-bader-ginsburg-counsel — the long game and the test case
   • Digital-platform review: lawrence-lessig-counsel + tim-wu-counsel — code is law meets the Cycle

5. Dispatch the panel in parallel via the Agent tool. Each persona gets:

   • The full target content (file or directory contents)
   • The bible context (CLAUDE.md, prior memos, prior reviews, the underlying decision)
   • Specific instruction to produce a review in their voice with: Verdict (one sentence top-line), Marked passages (3-8 quoted excerpts with strikethroughs for cuts and [→ replacement] for substitutions), Hand-off (if a different persona would serve better).

6. Consolidate the parallel returns. The output is a single review file:

---
title: Counsel review of <path>
slug: <slug>
panel: [<persona-1>, <persona-2>, <persona-3>]
created: YYYY-MM-DD
target: <path>
target_lines: <N>
---

# Counsel review: <path>

> ⚠️ **Craft register only — not legal advice.** This review is a reasoning exercise from three counsel lenses. It is not a representation by counsel. Real legal questions require real lawyers.

## Per-persona verdicts

### <persona-1>
**Verdict:** <one sentence top-line reaction>

**Marked passages:**
- <quoted excerpt 1>
- <quoted excerpt 2>
- ...

**Hand-off:** <if a different persona would serve better; or omit>

### <persona-2>
[same structure]

### <persona-3>
[same structure]

## Where they agree

<1-3 points where the panel converges. The strongest signal — when independent reviewers with different lenses flag the same thing, it's almost certainly real.>

## Where they disagree

<1-2 points where the panel diverges. Often the most useful section — disagreement reveals the genuine trade-off the author is making. RBG vs. Scalia disagreements in particular are load-bearing, not noise.>

## Highest-leverage change

<ONE recommendation. The single change that, if made, would make the artifact meaningfully better. Not a list. The orchestrator's job is to pick the highest-leverage move; this skill picks it for them.>

## Suggested next step

<One of: implement the change, escalate to a real attorney, request a v2 from the memo author, run /counsels-debate <topic> <persona-A> <persona-B> if the disagreement is structural (RBG vs. Scalia is the canonical example), cross-dispatch great-minds:marcus-aurelius-mod if the question is interpersonal/Stoic-mediation-shaped.>

7. Save the review to counsel/reviews/<slug>.md. Slug derived from the target path or from CLAUDE.md's Current memo: field.

8. Report:

   📝 Saved to counsel/reviews/<slug>.md (review of <path>, <word-count> words).
   ⚠️ Craft register only — not legal advice.
   
   Panel:           <persona-1>, <persona-2>, <persona-3>
   Convergence:     <one-line summary of the agreement>
   Highest-leverage change: <one-line summary>
   
   Next:
   - Address the highest-leverage change
   - Or run /counsels-debate <topic> <a> <b> if the disagreement is structural
   - For real legal stakes: retain a licensed attorney
   

What the skill does NOT do

• Modify the target. It reviews; the author (human or AI) edits.
• Approve or reject. It surfaces the review's findings; the orchestrator decides.
• Provide legal representation. The personas are craft channels, not licensed counsel.
• Auto-pick the panel based on opaque heuristics. The default panel is documented; alternatives are documented; the user can override explicitly.

Notes

• Parallel dispatch means the personas don't see each other's reviews when forming their own. This is intentional — independent verdicts produce stronger signal at consolidation time.
• For very large targets (>2,000 lines, or a whole policy library), the dispatch brief should ask each persona to focus on the most-critical sections only — pick a sample, don't try to review the whole thing. Better to do three high-quality reviews of focused sections than three shallow reviews of everything.
• Review fatigue is real. Don't run /counsels-review on every memo; use it for non-trivial decisions where the consolidated lens is worth the dispatch cost.
• The disclaimer block is not optional. Do not strip the "Craft register only — not legal advice" line from any review.