adversarial-review

Three parallel reviewer agents find contradictions, config gaps, and coverage holes across a project's docs, config, and tests — then fix what you approve. Heavyweight (dispatches parallel agents, can modify files). Supports env-var automation hooks (ADVERSARIAL_REVIEW_TARGETS, ADVERSARIAL_REVIEW_REPORT_ONLY) so an orchestrator can dispatch a programmatic validator pass without keyboard. Use only when explicitly asked: /adversarial-review, "adversarial review", "find gaps in my docs/config", OR when another skill (e.g. /feature) needs to dispatch a validator pass programmatically.

Quality

75%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

High

Do not use without reviewing

Fix and improve this skill with Tessl

tessl review fix ./plugins/agentic-engineering/skills/adversarial-review/SKILL.md

Quality

Content

63%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a well-sequenced, actionable orchestration playbook with real validation gates, but it is verbose and monolithic — repeated file lists and inlined agent prompts inflate the token budget and skip progressive disclosure entirely.

Suggestions

Move the three full reviewer-agent prompt templates (spec/config/coverage) into a references/ file and reference them by path, cutting the inline bulk and enabling one-level-deep progressive disclosure.

Define the project-instruction-file list once and reference it by name instead of repeating the full CLAUDE.md/AGENTS.md/GEMINI.md/... enumeration three times.

Trim the discovery-step explanations of globbing and CI-config conventions that Claude already knows, keeping only the project-specific file-glob lists.

Dimension	Reasoning	Score
Conciseness	Mostly efficient but padded in places — the project-instruction file list (CLAUDE.md, AGENTS.md, GEMINI.md, ...) is repeated verbatim three times, and several passages over-explain globbing and CI conventions Claude already knows.	3 / 5
Actionability	Provides concrete, executable guidance — specific git commands ('git diff <base>..HEAD --name-status'), named env vars, and copy-paste agent-prompt templates — with only minor gaps from intentional [PLACEHOLDER] template directives.	4 / 5
Workflow Clarity	Clear Phase 1–5 sequence with an explicit user-approval checkpoint (Phase 3) gating destructive edits and a retry-once loop for failed agents; post-fix verification is only advisory ('suggest running again'), the minor gap that keeps it below 5.	4 / 5
Progressive Disclosure	Has usable phase structure, but the 308-line body inlines three long agent-prompt templates and the full discovery checklist rather than offloading them to reference files; no bundle files exist and no one-level-deep references are signaled.	3 / 5
	Total	14 / 20 Passed

Description

87%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is strong: it states concrete capabilities, gives explicit natural-language triggers, and scopes usage to avoid conflicts. Minor specificity/trigger-synonym gaps keep two dimensions at 4 rather than 5.

Dimension	Reasoning	Score
Specificity	Names the domain and several concrete actions — 'find contradictions, config gaps, and coverage holes', 'fix what you approve', 'dispatches parallel agents' — but stops short of the fully comprehensive enumeration of the 5 anchor.	4 / 5
Completeness	Explicitly answers both what ('find contradictions, config gaps, and coverage holes... then fix what you approve') and when ('Use only when explicitly asked: /adversarial-review, "adversarial review", "find gaps in my docs/config"...') with concrete trigger phrases.	5 / 5
Trigger Term Quality	Includes natural trigger phrases a user would actually say — '/adversarial-review', 'adversarial review', 'find gaps in my docs/config' — with good keyword coverage, though it lacks the synonym/extension breadth of the 5 anchor.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (parallel adversarial doc/config/test review) with distinct triggers, and the 'Use only when explicitly asked' guard actively minimizes wrong-skill conflict.	5 / 5
	Total	18 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: belchman/claude-skills
Path: plugins/agentic-engineering/skills/adversarial-review/SKILL.md
Commit: 77a9e6b

Reviewed: 11 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.