# AgentTaste skill directory
> Browse skills by task. Upstream entries link to verified source revisions; AgentTaste entries contain authored instructions.

JSON catalog: https://agenttaste.dev/skills.json?access=free&category=development
Human directory: https://agenttaste.dev/skills?access=free&category=development

## How to choose and use a skill
Choose and install the appropriate skills in my current project, then use them to complete the task I describe next.
Read https://agenttaste.dev/skills.json?access=free&category=development.
Choose skills by their useWhen, inputs, prerequisites and expected outputs.
Read the full SKILL.md and any referenced upstream files before using a skill.
Explain the chosen skills and check that their prerequisites and access terms are met.
Use the installation instructions for the agent running this task.
Use the listed project installation command or the documented plugin setup. Inspect existing installed skills before modifying them, and keep companion rules and scripts together.
Do not invent packages, search metrics, credentials or compatibility. If a required tool is unavailable, say which prerequisite is missing.
Then apply the chosen skill to my task and verify the result against the project instructions.

## Engineering (development)
Architecture, TypeScript rules and React performance.

### Cursor Rules Strict TypeScript Architect (cursor-rules-architect)
Prevents Cursor from taking lazy shortcuts: forbids type assertions (as any), enforces readonly properties on domain models, and checks for unused imports.

#### Use when
- Enforce TypeScript and server-boundary rules across a project.

#### Required inputs
- TypeScript repository
- Framework and validation requirements

#### Prerequisites
- Access to the compiler and relevant project checks.

#### Expected outputs
- Reusable components and explicit input validation

Source: https://agenttaste.dev/skills/cursor-rules-architect#instructions
Full SKILL.md: https://agenttaste.dev/skills/cursor-rules-architect/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Cursor, Claude Code, Codex
Installation (Claude Code):
```text
curl --fail --show-error --location --create-dirs 'https://agenttaste.dev/skills/cursor-rules-architect/SKILL.md' --output '.claude/skills/cursor-rules-architect/SKILL.md'
```

### React best practices (vercel-react-best-practices)
Vercel Engineering’s React performance skill includes a companion rule collection. Install the package so the referenced rules remain available.

#### Use when
- Write, review or refactor React or Next.js code.
- Investigate unnecessary rendering or slow data loading.

#### Required inputs
- React or Next.js repository
- Performance symptoms or implementation task

#### Prerequisites
- React or Next.js project.
- Install the complete upstream skill directory, including its rules.

#### Expected outputs
- Prioritized performance changes
- Validation against the relevant rules

Source: https://github.com/vercel-labs/agent-skills/blob/063bee94c3f4df8453406c830b0a7df0f2860278/skills/react-best-practices/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/vercel-react-best-practices/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/vercel-labs/agent-skills/tree/063bee94c3f4df8453406c830b0a7df0f2860278/skills/react-best-practices --skill vercel-react-best-practices --agent claude-code
```

### Web app testing with Playwright (anthropic-webapp-testing)
Anthropic’s webapp-testing toolkit uses Python Playwright and includes a server-lifecycle helper, examples and browser-log workflows for testing local applications.

#### Use when
- Verify a local web app’s interactions or reproduce a frontend bug.

#### Required inputs
- Local application URL or startup commands
- User journeys and expected behavior

#### Prerequisites
- Python with Playwright and its browser binaries installed.
- A locally runnable application; install the entire skill directory to retain scripts and examples.

#### Expected outputs
- Browser test scripts
- Screenshots, logs and observed test results

Source: https://github.com/anthropics/skills/blob/dbd4588f9e1033efb41dad4bef2f7947c8993d44/skills/webapp-testing/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/anthropic-webapp-testing/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/anthropics/skills/tree/dbd4588f9e1033efb41dad4bef2f7947c8993d44/skills/webapp-testing --skill webapp-testing --agent claude-code
```

### MCP server builder (anthropic-mcp-builder)
Anthropic’s MCP development guide covers protocol research, Python or TypeScript implementation, tool design, testing and task-based evaluations, with companion references and scripts.

#### Use when
- Expose an external API or service through an MCP server.

#### Required inputs
- Service API documentation and allowed operations
- Target MCP client, language and authentication requirements

#### Prerequisites
- Python or Node.js development environment and the appropriate MCP SDK.
- Access to current protocol documentation; use test credentials for authenticated integrations.

#### Expected outputs
- MCP server implementation
- Tool tests and evaluation questions

Source: https://github.com/anthropics/skills/blob/dbd4588f9e1033efb41dad4bef2f7947c8993d44/skills/mcp-builder/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/anthropic-mcp-builder/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/anthropics/skills/tree/dbd4588f9e1033efb41dad4bef2f7947c8993d44/skills/mcp-builder --skill mcp-builder --agent claude-code
```

### Skill authoring and evaluation (anthropic-skill-creator)
Anthropic’s skill-creator workflow iterates on skill instructions, trigger descriptions and evaluations. The bundled evaluation workflow uses Claude tooling; install the scripts and reference material together.

#### Use when
- Create a reusable agent skill or evaluate an existing skill.

#### Required inputs
- Task definition and representative prompts
- Existing skill, expected outputs and evaluation criteria

#### Prerequisites
- Claude Code plus Python and the bundled evaluation scripts for the documented evaluation workflow.
- Model calls used by evaluations may incur usage charges.

#### Expected outputs
- Versioned SKILL.md package
- Evaluation results and revised trigger description

Source: https://github.com/anthropics/skills/blob/dbd4588f9e1033efb41dad4bef2f7947c8993d44/skills/skill-creator/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/anthropic-skill-creator/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code
Installation (Claude Code):
```text
npx skills add https://github.com/anthropics/skills/tree/dbd4588f9e1033efb41dad4bef2f7947c8993d44/skills/skill-creator --skill skill-creator --agent claude-code
```

### React composition patterns (vercel-composition-patterns)
Vercel’s composition guide covers compound components, state ownership, context and React 19 APIs. Its companion rule files provide concrete component architecture examples.

#### Use when
- Refactor a component API with many boolean props.
- Design a reusable React component library.

#### Required inputs
- React components and their consumers
- Required variants, state ownership and composition constraints

#### Prerequisites
- React project source and access to all affected consumers.
- Install the complete upstream rule directory.

#### Expected outputs
- Composable component APIs
- Refactored consumers and relevant checks

Source: https://github.com/vercel-labs/agent-skills/blob/063bee94c3f4df8453406c830b0a7df0f2860278/skills/composition-patterns/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/vercel-composition-patterns/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/vercel-labs/agent-skills/tree/063bee94c3f4df8453406c830b0a7df0f2860278/skills/composition-patterns --skill vercel-composition-patterns --agent claude-code
```

### React Native performance practices (vercel-react-native-skills)
Vercel’s React Native collection organizes performance and implementation rules for Expo and React Native, including list virtualization, animation, navigation and monorepo dependencies.

#### Use when
- Build or review a React Native or Expo app.
- Investigate slow scrolling, rendering or mobile animations.

#### Required inputs
- React Native repository and target devices
- Performance symptoms or implementation task

#### Prerequisites
- React Native or Expo development environment.
- A simulator or physical device for runtime checks; retain the companion rules.

#### Expected outputs
- Prioritized mobile implementation changes
- Device validation checklist

Source: https://github.com/vercel-labs/agent-skills/blob/063bee94c3f4df8453406c830b0a7df0f2860278/skills/react-native-skills/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/vercel-react-native-skills/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/vercel-labs/agent-skills/tree/063bee94c3f4df8453406c830b0a7df0f2860278/skills/react-native-skills --skill vercel-react-native-skills --agent claude-code
```

### Postgres performance and security (supabase-postgres-best-practices)
Supabase’s Postgres guide supplies rules and SQL examples for query performance, connections, security, schema design and locking. It applies to Postgres beyond Supabase-hosted projects.

#### Use when
- Write or review Postgres queries, migrations or security policies.
- Diagnose slow queries or connection exhaustion.

#### Required inputs
- Schema, queries and query plans
- Postgres version and workload constraints

#### Prerequisites
- A Postgres project and representative queries or schema.
- A test database for executing changes; keep the referenced rule collection.

#### Expected outputs
- SQL or schema recommendations
- Relevant performance and access-control checks

Source: https://github.com/supabase/agent-skills/blob/c9be0e931b7930f7d02126d04774d904c381e7d7/skills/supabase-postgres-best-practices/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/supabase-postgres-best-practices/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/supabase/agent-skills/tree/c9be0e931b7930f7d02126d04774d904c381e7d7/skills/supabase-postgres-best-practices --skill supabase-postgres-best-practices --agent claude-code
```

### Cloudflare Agents SDK (cloudflare-agents-sdk)
Cloudflare’s Agents SDK skill routes implementation work to current official documentation and companion references for agent lifecycle, durable state, callable methods, scheduling and chat.

#### Use when
- Build or review an application using Cloudflare’s agents package.

#### Required inputs
- Workers repository and installed agents version
- State, scheduling and communication requirements

#### Prerequisites
- Cloudflare Workers project with the agents package and current generated types.
- Network access for documentation; Cloudflare account access is required for deployment.

#### Expected outputs
- Agent implementation and bindings
- Runtime and integration validation plan

Source: https://github.com/cloudflare/skills/blob/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/agents-sdk/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/cloudflare-agents-sdk/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/cloudflare/skills/tree/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/agents-sdk --skill agents-sdk --agent claude-code
```

### Cloudflare Durable Objects (cloudflare-durable-objects)
Cloudflare’s Durable Objects guide covers stateful coordination, authorization, stub creation, storage, alarms and tests, with current platform documentation as the API authority.

#### Use when
- Implement chat rooms, coordinated sessions or other stateful Workers features.

#### Required inputs
- Workers repository and coordination requirements
- Storage, identity and concurrency constraints

#### Prerequisites
- Workers project with Durable Object bindings and migration configuration.
- Local runtime or a permitted Cloudflare test environment for verification.

#### Expected outputs
- Durable Object implementation and configuration
- Storage, authorization and lifecycle tests

Source: https://github.com/cloudflare/skills/blob/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/durable-objects/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/cloudflare-durable-objects/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/cloudflare/skills/tree/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/durable-objects --skill durable-objects --agent claude-code
```

### Wrangler CLI workflows (cloudflare-wrangler)
Cloudflare’s Wrangler skill inspects existing Wrangler configuration, retrieves current command documentation and distinguishes local resources from remote changes before applying operations.

#### Use when
- Run or debug Wrangler in an existing configured project.

#### Required inputs
- wrangler.jsonc, wrangler.json or wrangler.toml
- Installed CLI version and target environment

#### Prerequisites
- An existing Wrangler-configured project and its local CLI dependency.
- Cloudflare credentials and explicit target selection for remote operations; new cf-based projects use Cloudflare’s separate cf-cli skill.

#### Expected outputs
- Verified CLI commands or configuration changes
- Validation of the intended Worker and resources

Source: https://github.com/cloudflare/skills/blob/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/wrangler/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/cloudflare-wrangler/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/cloudflare/skills/tree/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/wrangler --skill wrangler --agent claude-code
```

### Workers production practices (cloudflare-workers-best-practices)
Cloudflare’s Workers guide checks production code and configuration against installed versions, generated types and current documentation, including secrets, logs, streaming and runtime anti-patterns.

#### Use when
- Write or review a Cloudflare Worker before deployment.

#### Required inputs
- Worker source and configuration
- Runtime versions and production requirements

#### Prerequisites
- Workers repository with its installed toolchain and generated binding types.
- Network access to current Cloudflare documentation.

#### Expected outputs
- Specific code and configuration changes
- Type, build and runtime checks

Source: https://github.com/cloudflare/skills/blob/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/workers-best-practices/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/cloudflare-workers-best-practices/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/cloudflare/skills/tree/86505e44f87d801aa1864a3c4109e9a8ae5c409f/skills/workers-best-practices --skill workers-best-practices --agent claude-code
```

### Expo Router navigation (expo-router-navigation)
Expo’s routing skill covers route organization, dynamic segments, native stacks and tabs, headers, link previews and modal presentation, with SDK-specific reference files.

#### Use when
- Implement or restructure navigation in an Expo Router app.

#### Required inputs
- Expo repository and SDK version
- Route map and navigation behavior

#### Prerequisites
- Expo Router project and an appropriate simulator or device.
- Keep companion references and check API support against the installed SDK.

#### Expected outputs
- Route structure and navigation components
- Navigation and deep-link checks

Source: https://github.com/expo/skills/blob/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/expo-router/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/expo-router-navigation/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/expo/skills/tree/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/expo-router --skill expo-router --agent claude-code
```

### Expo data fetching and offline states (expo-data-fetching)
Expo’s networking guide covers fetch, React Query, SWR and router data loaders, including request debugging, caching and explicit screen states.

#### Use when
- Implement or debug network requests in an Expo application.

#### Required inputs
- Expo screen or data layer
- API contract, authentication and caching requirements

#### Prerequisites
- Expo application and an accessible API or test fixture.
- Check library and router-loader availability for the project’s installed versions.

#### Expected outputs
- Data-fetching implementation
- Explicit screen states and request validation

Source: https://github.com/expo/skills/blob/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/expo-data-fetching/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/expo-data-fetching/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/expo/skills/tree/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/expo-data-fetching --skill expo-data-fetching --agent claude-code
```

### Expo SDK upgrades (expo-sdk-upgrades)
Expo’s upgrade skill follows the current SDK migration process, dependency checks and native changes, with references for React, navigation, media packages and the New Architecture.

#### Use when
- Upgrade an Expo SDK or investigate upgrade-related build failures.

#### Required inputs
- Current SDK, dependency versions and native configuration
- Target SDK and observed build errors

#### Prerequisites
- Expo repository, package manager and a reproducible development build.
- Native toolchains or EAS access are needed for the corresponding build checks.

#### Expected outputs
- Updated dependencies and migration changes
- Post-upgrade build and device checks

Source: https://github.com/expo/skills/blob/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/expo-upgrade/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/expo-sdk-upgrades/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/expo/skills/tree/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/expo-upgrade --skill expo-upgrade --agent claude-code
```

### Expo EAS CI/CD workflows (expo-eas-workflows)
Expo’s EAS workflow skill covers workflow files, expressions and validation against current service documentation. The instructions are free; executing jobs uses EAS services and may incur charges.

#### Use when
- Create or review CI/CD automation for an Expo project.

#### Required inputs
- Expo project and eas.json
- Build, submission or deployment pipeline requirements

#### Prerequisites
- Expo account, EAS configuration and the documented CLI tooling.
- EAS plan limits and usage charges apply; store submission also requires the relevant developer accounts.

#### Expected outputs
- .eas/workflows YAML configuration
- Validation results and operational prerequisites

Source: https://github.com/expo/skills/blob/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/eas-workflows/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/expo-eas-workflows/SKILL.md
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/expo/skills/tree/7ddc1f5796996e5c7efe2fc3fa9168eb24e8dac1/plugins/expo/skills/eas-workflows --skill eas-workflows --agent claude-code
```

### Superpowers brainstorming (superpowers-brainstorming)
Jesse Vincent’s Superpowers brainstorming workflow uses focused questions and design discussion to turn an uncertain request into an agreed direction before implementation.

#### Use when
- Explore a feature or substantial change whose requirements are still uncertain.

#### Required inputs
- Initial idea and intended users
- Repository context, constraints and success criteria

#### Prerequisites
- Access to the relevant project and a user who can settle product decisions.
- Install the complete skill directory; the optional visual companion uses a local Node.js server.

#### Expected outputs
- Clarified requirements and design decisions
- An agreed design ready for implementation planning

Source: https://github.com/obra/superpowers/blob/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/brainstorming/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/superpowers-brainstorming/SKILL.md

#### Popularity evidence
- 302,164 GitHub repo stars — obra/superpowers; checked 2026-10-20. Source: https://github.com/obra/superpowers/stargazers
- 398,168 skill installs (all time) — obra/superpowers / brainstorming · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/obra/superpowers/brainstorming
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/obra/superpowers/tree/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/brainstorming --skill brainstorming --agent claude-code
```

### Superpowers systematic debugging (superpowers-systematic-debugging)
Superpowers’ debugging discipline separates root-cause investigation, pattern analysis, hypothesis testing and implementation, with companion debugging techniques in the original directory.

#### Use when
- Diagnose a failing test, build, integration or production behavior.

#### Required inputs
- Failure symptoms and reproduction steps
- Relevant code, logs and execution environment

#### Prerequisites
- A reproducible failure or access to an environment that exhibits it.
- Retain the companion debugging references; do not share credentials in captured logs.

#### Expected outputs
- Evidence-backed root cause
- Targeted fix and regression verification

Source: https://github.com/obra/superpowers/blob/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/systematic-debugging/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/superpowers-systematic-debugging/SKILL.md

#### Popularity evidence
- 302,164 GitHub repo stars — obra/superpowers; checked 2026-10-20. Source: https://github.com/obra/superpowers/stargazers
- 300,287 skill installs (all time) — obra/superpowers / systematic-debugging · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/obra/superpowers/systematic-debugging
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/obra/superpowers/tree/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/systematic-debugging --skill systematic-debugging --agent claude-code
```

### Superpowers implementation plans (superpowers-writing-plans)
Superpowers’ planning skill writes a handoff for an engineer who has not seen the project, specifying interfaces, files, test commands and task boundaries from an already agreed design.

#### Use when
- Plan a multi-step implementation after requirements have been agreed.

#### Required inputs
- Approved specification or design
- Repository structure and existing test commands

#### Prerequisites
- An agreed specification and access to the codebase.
- Install the corresponding Superpowers execution skills separately for the documented execution handoff.

#### Expected outputs
- Versioned implementation plan
- Task-level acceptance and verification steps

Source: https://github.com/obra/superpowers/blob/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/writing-plans/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/superpowers-writing-plans/SKILL.md

#### Popularity evidence
- 302,164 GitHub repo stars — obra/superpowers; checked 2026-10-20. Source: https://github.com/obra/superpowers/stargazers
- 278,266 skill installs (all time) — obra/superpowers / writing-plans · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/obra/superpowers/writing-plans
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/obra/superpowers/tree/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/writing-plans --skill writing-plans --agent claude-code
```

### Superpowers test-driven development (superpowers-test-driven-development)
Superpowers’ TDD skill follows a red-green-refactor cycle, explicitly checking that a new test fails for the intended reason before implementing the behavior.

#### Use when
- Implement a feature or bug fix with a test-first workflow.

#### Required inputs
- Behavior to implement and expected result
- Project source and a runnable test suite

#### Prerequisites
- A working test runner and a testable interface for the requested behavior.
- Keep the bundled testing guidance with SKILL.md.

#### Expected outputs
- Behavior-focused regression tests
- Implementation with observed red and green results

Source: https://github.com/obra/superpowers/blob/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/test-driven-development/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/superpowers-test-driven-development/SKILL.md

#### Popularity evidence
- 302,164 GitHub repo stars — obra/superpowers; checked 2026-10-20. Source: https://github.com/obra/superpowers/stargazers
- 254,722 skill installs (all time) — obra/superpowers / test-driven-development · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/obra/superpowers/test-driven-development
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/obra/superpowers/tree/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/test-driven-development --skill test-driven-development --agent claude-code
```

### Superpowers completion verification (superpowers-verification)
Superpowers’ verification skill checks which command proves a claim, runs it, reads its full result and reports the result with its actual limits before declaring work complete.

#### Use when
- Verify a proposed completion, fix, build or test claim.

#### Required inputs
- Change or claim to verify
- Relevant acceptance criteria and validation commands

#### Prerequisites
- Ability to run the commands that demonstrate the claimed behavior.
- A passing check only supports the behavior that check actually covers.

#### Expected outputs
- Recorded verification results
- Accurate completion status and remaining limitations

Source: https://github.com/obra/superpowers/blob/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/verification-before-completion/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/superpowers-verification/SKILL.md

#### Popularity evidence
- 302,164 GitHub repo stars — obra/superpowers; checked 2026-10-20. Source: https://github.com/obra/superpowers/stargazers
- 246,086 skill installs (all time) — obra/superpowers / verification-before-completion · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/obra/superpowers/verification-before-completion
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/obra/superpowers/tree/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/verification-before-completion --skill verification-before-completion --agent claude-code
```

### Matt Pocock decision interview (matt-pocock-grilling)
Matt Pocock’s grilling skill models an idea as a tree of dependent decisions, asks the currently answerable questions and revisits the remaining branches as the user responds.

#### Use when
- Stress-test an implementation plan, design or product decision.

#### Required inputs
- Plan or idea to examine
- Known constraints and decision owners

#### Prerequisites
- A user available to make the decisions.
- The documented environment-research workflow uses an agent client with subagent dispatch.

#### Expected outputs
- Resolved decision tree
- Shared understanding of the proposed work

Source: https://github.com/mattpocock/skills/blob/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/productivity/grilling/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/matt-pocock-grilling/SKILL.md

#### Popularity evidence
- 297,596 GitHub repo stars — mattpocock/skills; checked 2026-10-20. Source: https://github.com/mattpocock/skills/stargazers
- 948,515 skill installs (all time) — mattpocock/skills / grilling · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/mattpocock/skills/grilling
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/mattpocock/skills/tree/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/productivity/grilling --skill grilling --agent claude-code
```

### Matt Pocock deep module design (matt-pocock-codebase-design)
Matt Pocock’s codebase-design reference defines modules, interfaces, test seams and adapters, then uses depth, leverage and locality to evaluate architecture and testability.

#### Use when
- Design a module interface or assess a refactor’s effect on testability.

#### Required inputs
- Current module and its consumers
- Behavior, dependency variation and test constraints

#### Prerequisites
- Project source and representative consumers.
- Keep DEEPENING.md and DESIGN-IT-TWICE.md; the optional alternative-design exercise needs subagent support.

#### Expected outputs
- Proposed interfaces and test seams
- Architecture tradeoffs grounded in callers

Source: https://github.com/mattpocock/skills/blob/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/codebase-design/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/matt-pocock-codebase-design/SKILL.md

#### Popularity evidence
- 297,596 GitHub repo stars — mattpocock/skills; checked 2026-10-20. Source: https://github.com/mattpocock/skills/stargazers
- 825,720 skill installs (all time) — mattpocock/skills / codebase-design · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/mattpocock/skills/codebase-design
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/mattpocock/skills/tree/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/codebase-design --skill codebase-design --agent claude-code
```

### Matt Pocock behavioral TDD (matt-pocock-tdd)
Matt Pocock’s TDD reference emphasizes tests through agreed public interfaces, useful assertions, narrow mocking and one behavior slice at a time.

#### Use when
- Implement a feature or fix while protecting behavior through public interfaces.

#### Required inputs
- Expected behavior and public interfaces
- Project test tools and source

#### Prerequisites
- A runnable test suite and agreement on which public interfaces to test.
- Install Matt Pocock’s codebase-design skill for its referenced interface and seam vocabulary.

#### Expected outputs
- Behavioral tests at agreed seams
- Incremental implementation and verification

Source: https://github.com/mattpocock/skills/blob/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/tdd/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/matt-pocock-tdd/SKILL.md

#### Popularity evidence
- 297,596 GitHub repo stars — mattpocock/skills; checked 2026-10-20. Source: https://github.com/mattpocock/skills/stargazers
- 1,114,501 skill installs (all time) — mattpocock/skills / tdd · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/mattpocock/skills/tdd
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/mattpocock/skills/tree/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/tdd --skill tdd --agent claude-code
```

### Matt Pocock domain modeling (matt-pocock-domain-modeling)
Matt Pocock’s domain-modeling skill challenges ambiguous terminology and captures domain concepts in GLOSSARY.md and scoped architecture decision records.

#### Use when
- Clarify inconsistent business terms or document an evolving domain model.

#### Required inputs
- Product domain and existing terminology
- Repository glossary and architecture decisions

#### Prerequisites
- Access to the repository and domain knowledge from the decision owner.
- Existing glossary maps and ADRs determine where scoped decisions belong.

#### Expected outputs
- Updated domain glossary
- Recorded decisions and edge-case scenarios

Source: https://github.com/mattpocock/skills/blob/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/domain-modeling/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/matt-pocock-domain-modeling/SKILL.md

#### Popularity evidence
- 297,596 GitHub repo stars — mattpocock/skills; checked 2026-10-20. Source: https://github.com/mattpocock/skills/stargazers
- 850,687 skill installs (all time) — mattpocock/skills / domain-modeling · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/mattpocock/skills/domain-modeling
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/mattpocock/skills/tree/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/domain-modeling --skill domain-modeling --agent claude-code
```

### Matt Pocock bug diagnosis (matt-pocock-bug-diagnosis)
Matt Pocock’s diagnosis loop begins with a fast red-capable reproduction, then minimizes it, ranks hypotheses, instruments specific predictions and verifies the final fix.

#### Use when
- Investigate a difficult bug or performance regression.

#### Required inputs
- Exact reported symptom
- Code, environment and permitted diagnostic tools

#### Prerequisites
- Access to an environment or fixture that can reproduce the reported symptom.
- Redact secrets in commands, logs and captured artifacts; retain companion diagnostic scripts.

#### Expected outputs
- Minimal reproduction and root cause
- Fix, regression evidence and diagnostic cleanup

Source: https://github.com/mattpocock/skills/blob/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/diagnosing-bugs/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/matt-pocock-bug-diagnosis/SKILL.md

#### Popularity evidence
- 297,596 GitHub repo stars — mattpocock/skills; checked 2026-10-20. Source: https://github.com/mattpocock/skills/stargazers
- 810,548 skill installs (all time) — mattpocock/skills / diagnosing-bugs · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/mattpocock/skills/diagnosing-bugs
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/mattpocock/skills/tree/5beb35b040e3de97e54fa9e386d7e42d15c295c2/skills/engineering/diagnosing-bugs --skill diagnosing-bugs --agent claude-code
```

### Addy Osmani code quality review (addy-osmani-code-review)
Addy Osmani’s code-review skill evaluates a change across five quality dimensions and reports actionable findings in the context of the project’s conventions.

#### Use when
- Review a diff, pull request or completed implementation.

#### Required inputs
- Diff or commit range
- Project standards and intended behavior

#### Prerequisites
- The complete diff and enough surrounding source to understand its effects.
- Repository standards and product requirements remain the review context.

#### Expected outputs
- Prioritized review findings
- Quality assessment with relevant verification

Source: https://github.com/addyosmani/agent-skills/blob/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/code-review-and-quality/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/addy-osmani-code-review/SKILL.md

#### Popularity evidence
- 108,660 GitHub repo stars — addyosmani/agent-skills; checked 2026-10-20. Source: https://github.com/addyosmani/agent-skills/stargazers
- 67,892 skill installs (all time) — addyosmani/agent-skills / code-review-and-quality · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/addyosmani/agent-skills/code-review-and-quality
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/addyosmani/agent-skills/tree/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/code-review-and-quality --skill code-review-and-quality --agent claude-code
```

### Addy Osmani application hardening (addy-osmani-security-hardening)
Addy Osmani’s security skill covers web application trust boundaries, authentication, authorization, data handling, dependency findings and external integrations.

#### Use when
- Implement or review web features that handle untrusted data or user sessions.

#### Required inputs
- Relevant source and data flows
- Threat context and dependency findings

#### Prerequisites
- Authorized access to the application source and a permitted test environment.
- A skill-based review does not establish compliance or replace a security assessment.

#### Expected outputs
- Concrete hardening changes or findings
- Validation of affected security behavior

Source: https://github.com/addyosmani/agent-skills/blob/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/security-and-hardening/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/addy-osmani-security-hardening/SKILL.md

#### Popularity evidence
- 108,660 GitHub repo stars — addyosmani/agent-skills; checked 2026-10-20. Source: https://github.com/addyosmani/agent-skills/stargazers
- 64,634 skill installs (all time) — addyosmani/agent-skills / security-and-hardening · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/addyosmani/agent-skills/security-and-hardening
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/addyosmani/agent-skills/tree/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/security-and-hardening --skill security-and-hardening --agent claude-code
```

### Addy Osmani performance optimization (addy-osmani-performance)
Addy Osmani’s performance guide spans frontend delivery, server execution and database queries, with profiling and measured evidence before prioritizing a fix.

#### Use when
- Investigate measured slowness or a known performance requirement.

#### Required inputs
- Workload and performance symptoms
- Source, profiles and target performance budget

#### Prerequisites
- A reproducible workload and access to the relevant profiling tools.
- Field performance data needs its own measurement source; this skill provides no live metrics.

#### Expected outputs
- Evidence-backed optimization changes
- Comparable before-and-after measurements

Source: https://github.com/addyosmani/agent-skills/blob/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/performance-optimization/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/addy-osmani-performance/SKILL.md

#### Popularity evidence
- 108,660 GitHub repo stars — addyosmani/agent-skills; checked 2026-10-20. Source: https://github.com/addyosmani/agent-skills/stargazers
- 60,118 skill installs (all time) — addyosmani/agent-skills / performance-optimization · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/addyosmani/agent-skills/performance-optimization
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/addyosmani/agent-skills/tree/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/performance-optimization --skill performance-optimization --agent claude-code
```

### Addy Osmani documentation and ADRs (addy-osmani-documentation-adrs)
Addy Osmani’s documentation skill records architecture choices, public API changes and operational context so future maintainers can understand the reasoning behind the code.

#### Use when
- Record an architectural choice or document a changed public interface.

#### Required inputs
- Decision and relevant implementation
- Alternatives, constraints and existing documentation

#### Prerequisites
- Accurate project facts and access to the documentation location.
- Use the repository’s existing documentation format and ownership conventions.

#### Expected outputs
- Decision record or updated documentation
- Clear context, consequences and references

Source: https://github.com/addyosmani/agent-skills/blob/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/documentation-and-adrs/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/addy-osmani-documentation-adrs/SKILL.md

#### Popularity evidence
- 108,660 GitHub repo stars — addyosmani/agent-skills; checked 2026-10-20. Source: https://github.com/addyosmani/agent-skills/stargazers
- 58,973 skill installs (all time) — addyosmani/agent-skills / documentation-and-adrs · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/addyosmani/agent-skills/documentation-and-adrs
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/addyosmani/agent-skills/tree/aeac41d4d005999432bb7312f7ff06383a91d59d/skills/documentation-and-adrs --skill documentation-and-adrs --agent claude-code
```

### Vercel agent-browser automation (vercel-agent-browser)
Vercel’s agent-browser skill is a discovery entry for the browser automation CLI. Its current stub loads the actual usage guide from the installed CLI so the workflow matches that version.

#### Use when
- Automate or inspect a website in an authorized browser workflow.

#### Required inputs
- Target URL and authorized task
- Browser session and any required application access

#### Prerequisites
- Node.js 24 or later, the agent-browser CLI and its browser binaries.
- The pinned SKILL.md is a discovery stub: run agent-browser skills get core to load the actual workflow; follow upstream setup before use.

#### Expected outputs
- Browser task results
- Relevant screenshots or observed interaction evidence

Source: https://github.com/vercel-labs/agent-browser/blob/5ee8691339a5a002e4d7454eec95250253bc7419/skills/agent-browser/SKILL.md
Full SKILL.md: https://agenttaste.dev/skills/vercel-agent-browser/SKILL.md

#### Popularity evidence
- 44,278 GitHub repo stars — vercel-labs/agent-browser; checked 2026-10-20. Source: https://github.com/vercel-labs/agent-browser/stargazers
- 1,179,586 skill installs (all time) — vercel-labs/agent-browser / agent-browser · skills.sh · all time; checked 2026-10-20. Source: https://skills.sh/vercel-labs/agent-browser/agent-browser
- 7,102,587 npm package downloads — agent-browser · 2026-09-19 – 2026-10-18; checked 2026-10-20. Source: https://api.npmjs.org/downloads/point/2026-09-19:2026-10-18/agent-browser
Adoption counts are not user ratings or guarantees of quality.
Access: Public. Upstream provider usage may have separate costs.
Agents: Claude Code, Cursor, Codex, Antigravity
Installation (Claude Code):
```text
npx skills add https://github.com/vercel-labs/agent-browser/tree/5ee8691339a5a002e4d7454eec95250253bc7419/skills/agent-browser --skill agent-browser --agent claude-code
```