New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Vibe Devil Advocate Review

by ash1794 in ash1794/vibe-engineering

Challenges a recommendation, design, large change, or any artifact submitted for hard review (spec, proposal, policy, plan) across 5 dimensions (consistency, completeness, actionability, alignment…

MITAuto-check passedAgent Workflows

Install Vibe Devil Advocate Review

skills CLI
$ npx skills add ash1794/vibe-engineering --skill vibe-devil-advocate-review -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install ash1794/vibe-engineering vibe-devil-advocate-review --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/ash1794/vibe-engineering.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/vibe-engineering/skills/vibe-devil-advocate-review .claude/skills/vibe-devil-advocate-review && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
vibe-devil-advocate-review
GitHub stars
163
Token cost
~2.3k tokens
SKILL.md length
1,259 words
Files
1
Skills in repo
33
Repo updated
First seen
Licence
MIT

At a glance

Challenges a recommendation, design, large change, or any artifact submitted for hard review (spec, proposal, policy, plan) across 5 dimensions (consistency, completeness, actionability, alignment…

  • Works in 3 steps: A different model family — for example,… → A fresh subagent that gets only the… → Self-review as a last resort. Explicitly…
  • Agent Workflows work in your project
  • SKILL.md covers When to Use This Skill, When NOT to Use This Skill, Get an Independent Reviewer and Peg the Reviewer to an Expert, plus 4 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Vibe Devil Advocate Review is an agent skill from ash1794/vibe-engineering. Challenges a recommendation, design, large change, or any artifact submitted for hard review (spec, proposal, policy, plan) across 5 dimensions (consistency, completeness, actionability, alignment, risk), from the standards of a named senior expert in the artifact's domain and ideally from a fresh context or a different model. Searches assuming defects exist and reports only those that survive evidence. Panel mode runs several independent lenses in parallel on a release candidate, verifies every finding against…

Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows. The repository describes itself as: 33 engineering discipline skills for Claude Code, OpenAI Codex & Gemini CLI + a CLI for CI/CD enforcement. Extracted from real-world multi-agent system development. Born from… The licence is MIT.

When your agent uses it

  • Agent Workflows work in your project

Example prompts

  • “Use the vibe-devil-advocate-review skill to challenge a recommendation, design, large change, or any artifact submitted for hard review (spec…”
  • “/vibe-devil-advocate-review”

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

  1. A different model family — for example, have Codex (GPT) review Claude's work, or Claude review Gemini's. Different training produces…
  2. A fresh subagent that gets only the artifact and the stated goals, not the conversation that produced them.
  3. Self-review as a last resort. Explicitly assume the artifact is wrong and look for the evidence.

What it can do on your machine

Read from SKILL.md and the folder at commit 8f1d71b. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Vibe Devil Advocate Review loads about 2.3k tokens when it runs. Until then it costs about 195 tokens; SKILL.md has 1,259 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~195
When it runs · the whole SKILL.md, loaded when a task matches
~2.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from ash1794/vibe-engineering at commit 8f1d71b, republished under its MIT licence (© ash1794). 1,259 words, ~2,261 tokens.

Download SKILL.mdSave it as .claude/skills/vibe-devil-advocate-review/SKILL.md (or your agent's skills folder).
name
vibe-devil-advocate-review
description
Challenges a recommendation, design, large change, or any artifact submitted for hard review (spec, proposal, policy, plan) across 5 dimensions (consistency, completeness, actionability, alignment, risk), from the standards of a named senior expert in the artifact's domain and ideally from a fresh context or a different model. Searches assuming defects exist and reports only those that survive evidence. Panel mode runs several independent lenses in parallel on a release candidate, verifies every finding against the current head, and routes confirmed ones to file owners. Use before shipping a significant recommendation, design, large branch, or multi-agent release, or when asked to tear something apart, be brutally honest, or poke holes in it.
user-invocable
true

vibe-devil-advocate-review

Before shipping a recommendation, challenge it. Models still tend to agree with their own earlier reasoning and with the user, and to read a capable-looking artifact as good. A reviewer that shares the author's context inherits the author's blind spots, so the review is strongest when the reviewer doesn't share them.

Two rules govern the whole review. Search like a pessimist: assume defects exist, and treat finding none as a sign the search wasn't hard enough yet. Report like a scientist: say only what survives as a real, evidenced defect. The first rule sets how hard you look; the second sets what you say. A manufactured finding is as damaging as flattery, because it teaches the author to discount the next real one.

When to Use This Skill

  • Before sending a design document for approval
  • Before shipping a recommendation that combines multiple inputs
  • Before merging a large feature branch
  • When you feel "too confident" about a solution
  • User asks for a review, a second opinion, or a brutally honest critique ("tear this apart", "poke holes in this", "what's wrong with this")
  • Reviewing a non-code artifact for hard critique: a proposal, policy, plan, curriculum, or argument
  • Reviewing your own draft before it ships

When NOT to Use This Skill

  • Trivial changes (typo fixes, formatting)
  • When the user explicitly says "just ship it"
  • During brainstorming (don't kill ideas before they form)
  • Small code changes (use vibe-quality-loop)
  • The user wants friendly proofreading or reassurance, not critique
  • Prose that reads as machine-written but is otherwise sound (use vibe-slop-filter)

Get an Independent Reviewer

In order of preference:

  1. A different model family — for example, have Codex (GPT) review Claude's work, or Claude review Gemini's. Different training produces different blind spots.
  2. A fresh subagent that gets only the artifact and the stated goals, not the conversation that produced them.
  3. Self-review as a last resort. Explicitly assume the artifact is wrong and look for the evidence.

Give the reviewer the artifact, the goals and constraints, the expert lens below, and this skill's 5 dimensions. Don't give it your own assessment.

Peg the Reviewer to an Expert

Skepticism without domain standards is contrarianism. Before writing a critical word, name the senior expert whose standards govern this artifact, and review from inside that person's judgment:

  • Code or architecture: a principal engineer who has maintained systems at scale and is unimpressed by code that works in the demo and fails in six months.
  • Design: a design lead who looks for the unspecified state, the edge case nobody drew, the accessibility gap.
  • Spec or requirements: the engineer who will have to build from it and test against it.
  • Policy or process: an institutional veteran who knows which clauses survive contact with real people and which become dead letters.
  • Proposal or argument: a referee in the field who spots the unsourced claim and the conclusion the evidence doesn't support.
  • Curriculum or training: a senior educator who sees missing scaffolding and assessments that don't measure the stated outcome.

If the artifact spans domains, name two or three lenses. If the right expert is genuinely unclear, ask one question first; reviewing from the wrong standard wastes the pass.

The expertise sits in the reviewer's chair; the charity is withheld from the author. Hold the work to the standard of the best in its field and treat every shortfall as a real defect, but don't invent errors or strawman it. Steelman what's there, then break it where it actually breaks.

The 5 Dimensions

  1. Consistency — Do all parts agree with each other? Any contradictions?
  2. Completeness — What's missing? Unaddressed edge cases? Blind spots? Is it 30% finished presenting as done?
  3. Actionability — Is every recommendation concrete and measurable? Could someone actually do it?
  4. Alignment — Does it match the stated goals, constraints, and user needs?
  5. Risk — What could go wrong? Second-order effects? Blast radius of failure?
Show full SKILL.md (614 more words)Show less

Steps

  1. Read the whole artifact before critiquing. Don't skim, and don't react to the first half. Many of the sharpest findings are cross-document: an objective in section 1 that the test plan in section 6 never measures, a claim made early and contradicted late.
  2. For each dimension, actively look for problems. Assume there are some.
  3. Order by what kills the artifact fastest: structural integrity first, then whether the central idea holds, then the domain-specific failures only the expert would catch, and cosmetic issues last. Typos matter mostly as tells about rigor elsewhere.
  4. Ask what the artifact actually is. Is it what it presents itself as? A research spike dressed as a deployment plan, a decision the author was meant to make quietly resolved by the template, another organization's voice on a problem that isn't theirs. This read often changes which findings matter.
  5. Verify each issue — Keep only issues you can support with evidence (a quote, file:line, or a concrete failure scenario). "The design is weak" is not a finding; "Section 3 names the retry budget as the goal but nothing defines how it's measured" is. Drop what you can't support.
  6. Name genuine strengths in one line each. Never manufacture a compliment, and don't dwell. If much of it is good, the review can be short.
  7. Score each dimension 1–5 (1 = critical issues, 5 = solid)
  8. Verdict: APPROVE / REVISE (with required changes) / REJECT (with blocking issues)
Gate the Review Before Sending

Run your own review through the same standard:

  • Did I find each issue, or need something to say? Strike anything that wouldn't survive the author asking "is that actually a problem?"
  • Is every issue located and evidenced?
  • Did I review from a real expert standard, or just lean negative?
  • Did I read the whole artifact?
  • If the work is good, did I say so plainly?

Three real defects and two named strengths beat ten findings, six of which are noise.

Delivery
  • Lead with the hardest finding. No "great work, but".
  • End when the findings end. No reassuring close.
  • If the user will add their own reflections afterwards, deliver the full review, then hand back explicitly and respond to what they add rather than repeating yourself.
  • Match depth to stakes: a document going to leadership gets a deeper read than a quick gut check.

Panel Mode (release candidates built by several agents)

One reviewer carries one set of blind spots, and unverified findings waste fix cycles. For a release or content lock:

  1. Freeze a head. Every lens reviews the same commit.
  2. Run lenses in parallel, each with a narrow brief and a bounded report format (file:line, severity). Pick lenses that fit the product, for example: editorial and tone, facts and privacy, UX and accessibility, engineering and performance.
  3. Verify separately. A distinct stage reproduces each finding on the current head and drops stale or false ones (already fixed, or a rule firing on text that already complies).
  4. Dedupe and rank P0–P2 across lenses.
  5. Route each confirmed finding to the owner of the file (vibe-workstream-orchestration), then re-run only the affected lenses.

If the harness supports scripted workflows, run the panel as one: it verifies more rigorously than routing findings by hand.

Output Format

Devil's Advocate Review

Reviewer: [different model / fresh subagent / self] Expert lens: [e.g. principal engineer, maintains payment systems at scale]

DimensionScoreIssues
ConsistencyX/5[count]
CompletenessX/5[count]
ActionabilityX/5[count]
AlignmentX/5[count]
RiskX/5[count]
Critical Issues
  1. [issue, location, evidence, concrete failure scenario]
Warnings
  1. [non-blocking concern, with location]
What Holds Up
  • [one line per genuine strength; omit the section if there are none]
Verdict: APPROVE / REVISE / REJECT

[rationale]

© ash1794, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in plugins/vibe-engineering/skills/vibe-devil-advocate-review of ash1794/vibe-engineering.

Open the folder on GitHubat commit 8f1d71b

Compare with similar skills

Vibe Devil Advocate Review next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Vibe Devil Advocate Review compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Vibe Devil Advocate Review this skillash1794/vibe-engineering163—~2.3kAutomated safety check: PassMIT
MCP Server Builderanthropics/skills180k63 repos~2.3kAutomated safety check: PassApache-2.0
Hook Development for Claude Code Pluginsanthropics/claude-plugins-official38k10 repos~4.1kAutomated safety check: NotesApache-2.0
Using Superpowersfarm-fe/farm5.6k36 repos~1.4kAutomated safety check: PassMIT
Executing Plans Inlineobra/superpowers297k2 repos~5.1kAutomated safety check: PassMIT
Skill CreatorAzure/azqr79689 repos~8.2kAutomated safety check: PassApache-2.0

Similar skills

  • MCP Server Builder

    anthropics/skills

    Official

    Guides the design and implementation of Model Context Protocol servers in TypeScript or Python, from tool naming and error messages to evaluation.

    180k GitHub starsUsed in 63 repos~2.3k tokens
    Agent WorkflowsAuto-check passed

  • Hook Development for Claude Code Plugins

    anthropics/claude-plugins-official

    Official

    Explains how to write Claude Code plugin hooks, both prompt-based checks and bash commands, for events such as PreToolUse, Stop and SessionStart.

    38k GitHub starsUsed in 10 repos~4.1k tokens
    Agent WorkflowsAuto-check: notes

  • Using Superpowers

    farm-fe/farm

    A skill your agent uses when starting any conversation - establishes how to find and use skills, requiring Skill tool invocation before ANY response including clarifying questions

    5.6k GitHub starsUsed in 36 repos~1.4k tokens
    Agent WorkflowsAuto-check passed

  • Executing Plans Inline

    obra/superpowers

    Has the agent carry out an implementation plan itself, task by task in the current session, keeping a ledger, proving each step with a test and ending with one whole-branch review.

    297k GitHub starsUsed in 2 repos~5.1k tokens
    Agent WorkflowsAuto-check passed

  • Skill Creator

    Azure/azqr

    Official

    Create new skills, modify and improve existing skills, and measure skill performance.

    796 GitHub starsUsed in 89 repos~8.2k tokens
    Agent WorkflowsAuto-check passed

  • Claude Code Agent Development

    anthropics/claude-plugins-official

    Official

    Explains how to write agents for Claude Code plugins: the markdown file with YAML frontmatter, trigger descriptions, model and color settings, and system prompt design.

    38k GitHub starsUsed in 7 repos~2.8k tokens
    Agent WorkflowsAuto-check passed

More from ash1794/vibe-engineering

All 33 skills in this repo

  • Vibe Concurrent Test Safety

    ash1794/vibe-engineering

    Audits tests for concurrency safety — race conditions, shared mock state, cleanup ordering.

    163 GitHub stars~665 tokensUpdated 3 days ago
    Auto-check passed

  • Vibe Fuzz Parser Inputs

    ash1794/vibe-engineering

    Generates fuzz test scaffolding for parsers handling external input (YAML, JSON, config files, user input).

    163 GitHub stars~722 tokensUpdated 3 days ago
    Auto-check passed

  • Vibe Golden File Testing

    ash1794/vibe-engineering

    Implements snapshot/golden file tests with temporal normalization so tests don't break daily.

    163 GitHub stars~669 tokensUpdated 3 days ago
    Auto-check passed

  • Vibe Parallel Task Decomposition

    ash1794/vibe-engineering

    Analyzes large tasks for independent subtasks that can be safely parallelized.

    163 GitHub stars~717 tokensUpdated 3 days ago
    Auto-check passed

  • Vibe Slop Filter

    ash1794/vibe-engineering

    Strips AI-generation "smell" from prose before it ships (READMEs, docs, release notes, PR descriptions, posts, emails).

    163 GitHub stars~2.3k tokensUpdated 3 days ago
    Auto-check passed

  • Vibe Spec Sync

    ash1794/vibe-engineering

    Keeps specification documents and code in agreement. An agent skill from ash1794/vibe-engineering.

    163 GitHub stars~2.2k tokensUpdated 3 days ago
    Auto-check passed

Categories

Questions about Vibe Devil Advocate Review

What does Vibe Devil Advocate Review do?

Challenges a recommendation, design, large change, or any artifact submitted for hard review (spec, proposal, policy, plan) across 5 dimensions (consistency, completeness, actionability, alignment…. Vibe Devil Advocate Review is an agent skill from ash1794/vibe-engineering. Challenges a recommendation, design, large change, or any artifact submitted for hard review (spec, proposal, policy, plan) across 5 dimensions (consistency, completeness, actionability, alignment, risk), from the standards of a named senior expert in the artifact's domain and ideally from a fresh context or a different model.

When should I use Vibe Devil Advocate Review?

Vibe Devil Advocate Review fits situations like: agent Workflows work in your project.

How do I install Vibe Devil Advocate Review in Claude Code?

Run `npx skills add ash1794/vibe-engineering --skill vibe-devil-advocate-review -a claude-code`. Or copy the skill folder (plugins/vibe-engineering/skills/vibe-devil-advocate-review in ash1794/vibe-engineering) into .claude/skills/vibe-devil-advocate-review in your project. Claude Code loads it when a task matches its description.

How do I install Vibe Devil Advocate Review in Codex?

Run `npx skills add ash1794/vibe-engineering --skill vibe-devil-advocate-review -a codex`. Or copy the skill folder (plugins/vibe-engineering/skills/vibe-devil-advocate-review in ash1794/vibe-engineering) into .agents/skills/vibe-devil-advocate-review in your project. Codex loads it when a task matches its description.

Can I use Vibe Devil Advocate Review in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add ash1794/vibe-engineering --skill vibe-devil-advocate-review -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/vibe-devil-advocate-review, .gemini/skills/vibe-devil-advocate-review, .github/skills/vibe-devil-advocate-review and .opencode/skills/vibe-devil-advocate-review in your project.

What does Vibe Devil Advocate Review need to run?

SKILL.md names no scripts, command-line tools or credentials: Vibe Devil Advocate Review is instructions for the agent only.

Does Vibe Devil Advocate Review access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Vibe Devil Advocate Review safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Vibe Devil Advocate Review use?

Vibe Devil Advocate Review is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Vibe Devil Advocate Review use?

About 2.3k tokens (SKILL.md is roughly 9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Vibe Devil Advocate Review?

Skills that share tags, products or a category with Vibe Devil Advocate Review: MCP Server Builder (anthropics/skills, 180k stars), Hook Development for Claude Code Plugins (anthropics/claude-plugins-official, 38k stars), Using Superpowers (farm-fe/farm, 5.6k stars) and Executing Plans Inline (obra/superpowers, 297k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Vibe Devil Advocate Review?

ash1794 (a GitHub user) maintains it in ash1794/vibe-engineering, which has 163 GitHub stars. The repository holds 33 skills in this directory. The repository was last updated on October 7, 2026.

Source: ash1794/vibe-engineering on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.