View on GitHubRecursive adversarial review skill for AI agents: discover bugs, trace root causes, fix them, verify, and re-attack.
About gauntlet
Gauntlet is an agent skill from aakashH242, under the MIT licence, for adversarial review and root-cause repair of code, designs, APIs, infrastructure and agent workflows. The README describes it as a recursive review skill that pushes an agent to look for counterexamples and back up its findings before saying a task is complete.
The workflow in the README runs from mapping, attacking and substantiating findings to tracing causes, repairing, verifying and attacking again within a finite budget. Reviews stay read-only unless fixes are explicitly authorised, and it can use sub-agents or separate solo passes. An optional Markdown ledger or bundled Python helper backed by SQLite tracks tasks and findings; the helper needs Python 3.10 and uses only the standard library. Codex and Claude Code are named as hosts. The README admits behaviour has not yet been evaluated across hosts or models. Install is through the skills command line tool or by copying the folder. These details come from the project's README.
Key features
- Recursive adversarial review
- Root-cause tracing and repair
- Read-only unless fixes authorised
- Optional SQLite ledger helper
- Works with sub-agents or solo passes
More coding & review agent skills
- Coding & review

Web Design Guidelines (Vercel)
vercel-labs
Review UI code for Web Interface Guidelines compliance. Use when asked to "review my UI", "check accessibility", "audit design", "review UX", or "check my site against best practices".
34K GitHub stars - Coding & review

Webapp Testing Skill
anthropics
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
195K GitHub stars - Coding & review

Web Artifacts Builder
anthropics
Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or s
195K GitHub stars - Coding & review

Vercel Optimize
vercel-labs
Use for Vercel cost and performance optimization on deployed projects, especially Next.js, SvelteKit, Nuxt, and limited Astro apps.
34K GitHub stars - Coding & review

Superpowers
obra
Superpowers is a community skill collection for Claude Code covering planning, test-driven development, debugging workflows and systematic problem solving.
318K GitHub starsShellMIT - Coding & review
Perform language and framework specific security best-practice reviews and suggest improvements. Trigger only when the user explicitly requests security best practices guidance, a security review/report, or secure-by-def
29K GitHub stars