adversarial-plan-review

Adversarial Plan Review

Pre-implementation review of a plan. The host (you, the agent reading this) must not review your own work — route the heavy critique to the other agent. This SKILL.md is the same in Claude Code and Codex; lib/call-external.sh detects which host you are and picks the opposite partner.

Cross-host principle

• You are running in Claude Code → external reviewer is Codex
• You are running in Codex → external reviewer is Claude (Opus, xhigh)
• All externals unavailable → DEGRADED MODE: host self-review with explicit banner. Never silently auto-review.

How to execute

1. Resolve the plan

If user gave plan text inline → use it. If user pointed to a file → use the Read tool on the file. If input is code or a diff → suggest /adversarial-review:coding-adversarial-review instead and stop here. If no input → ask: "What plan should I review? Paste it or point to a file."

2. Build the external-reviewer prompt

Use this template, replacing {PLAN_TEXT} with the actual plan:

You are an adversarial plan reviewer. Assume this plan will fail. Prove it.

PLAN:
{PLAN_TEXT}

Validate for:
1. Scope alignment — does the plan match stated objectives?
2. Missing steps — gaps in sequence (testing, migration, rollback)?
3. Dependency ordering — can steps execute as ordered? Circular deps?
4. Rollback strategy — what if step N fails? Reversible?
5. Blast radius — what existing functionality is at risk?
6. Success criteria — verifiable completion conditions?
7. Cost estimate — complexity, files changed, test impact

Output language: same as the input plan.
Sections: BLOCKERS / SHOULD FIX / NICE TO HAVE / VERDICT
Per finding: P0–P3 severity, evidence (line of plan), problem, recommendation.
Verdict: PROCEED / REVIEW_NEEDED / RETHINK
Provide an improved version of the plan incorporating the recommendations.

Keep the prompt focused. If the plan is over ~6 kB, summarize sections instead of pasting raw — long prompts can stall the external backend.

3. Call the external partner

Pipe the prompt into lib/call-external.sh (this script handles host detection, routing, anti-recursion, Gemini fallback, and degraded mode):

PLUGIN_DIR="$HOME/Documents/Repos/coding-plugins/adversarial-review"  # or wherever installed
echo "$PROMPT" | bash "$PLUGIN_DIR/lib/call-external.sh"
echo "exit=$?"

Capture:

• stdout = the partner's analysis (or degraded host-self if all failed)
• stderr = operational logs (which partner was used, latency, fallback chain)
• exit code = 0 external success, 2 degraded, 1 error/recursion

Notes:

• Do not call codex exec or claude -p directly — always go through lib/call-external.sh. The script enforces anti-recursion via the ADVERSARIAL_REVIEW_DEPTH env counter.
• If exit is 1 (recursion), you are inside a partner-launched call; emit a short note ("recursion guard tripped — parent already running review") and stop. Do not produce a self-review.

4. Run your own independent analysis (host-side)

Without looking at the partner's output, walk the same checklist (scope, missing steps, ordering, rollback, blast radius, success criteria). This is your host-side draft.

5. Cross-validate

Compare host-side findings with the partner's:

Tag	Meaning
[cross-validated]	both you and partner caught it (high confidence)
[external-only]	only the partner caught it
[host-only]	only you caught it

On severity disagreements, take the higher of the two.

6. Return unified output

Format:

## Adversarial Plan Review

- **Mode**: <external=codex | external=claude-opus | external=gemini-3.1-pro | gemini-3.1-flash-lite | gemini-2.5-pro | gemini-2.5-flash | DEGRADED>
- **Verdict**: PROCEED | REVIEW_NEEDED | RETHINK
- **Findings**: N total — X P0, Y P1, Z P2, W P3

### Critics

#### [P0]: <title>  [cross-validated | external-only | host-only]
- **Problem**: <what's wrong, why it matters>
- **Evidence**: <line of plan, quote>
- **Recommendation**: <specific fix>

[…repeat, highest severity first…]

### Revised Plan

<the complete improved plan, ready to execute — full text, not a diff>

### Key Changes from Original

1. <change> — <why>
2. <change> — <why>

If lib/call-external.sh exited 2 (degraded mode), prepend this banner verbatim to the top of the output, before the ## Adversarial Plan Review heading:

> ⚠️ **DEGRADED MODE** — no external partner reachable. Output below is
> single-perspective host self-review and violates the cross-host principle.
> Re-run after restoring access to Codex / Claude / Gemini for higher confidence.

References

• references/host-detection.md — how lib/detect-host.sh decides
• references/codex-integration.md — Codex CLI invocation, including the forced_login_method = "chatgpt" gotcha for ChatGPT-account auth
• references/claude-integration.md — claude -p --model opus --effort xhigh
• references/fallback-chain.md — full external + Gemini cascade + degraded path
• references/output-standards.md — P0–P3 schema, evidence requirements