DOCSEARCHbyAlgolia
Submit Docs

CodexSkills

repository·main·Indexed 21 days ago

https://github.com/am-will/codex-skills

A collection of specialized skills and hook bundles for AI agents like Codex and Claude. It provides automation hooks for planning, documentation access, frontend development, and browser automation, including tools such as agents-md-loader, build-on-change, change-logger, dependency-checker, deployment-health-monitor, and various notification integrations for Discord and Slack.

Tokens
77.2K
Snippets
185
Records
372
Agent score
76%

What's inside codex-skills

  1. Overview of the Codex Skill Expansion Plan

    main

    The Codex Skill Expansion Plan outlines a roadmap to fill gaps in the current skill library by adding high-impact capabilities for DevOps, automation, architecture, and AI/ML workflows. The goal is to provide complementary capabilities that allow for chained, modular operations (e.g., combining test-qa-orchestrator with ci-cd-pipeline-builder).

    Key focus areas include:

    • Git/GitHub workflow automation
    • CI/CD pipeline creation and troubleshooting
    • Testing automation and quality gates
    • Database migrations and schema management
    • Environment and secrets management
    • Code review and static analysis
    • API development and testing
    • Infrastructure scaffolding and verification
    • AI/ML workflow helpers (RAG, embeddings, prompt evaluation)

  2. Overview of Vercel React Best Practices

    main
    The vercel-react-best-practices skill provides a comprehensive set of 45 performance optimization rules for React and Next.js applications. These rules are categorized by impact and priority to guide automated refactoring, code generation, and manual code reviews. Use this skill when writing, reviewing, or refactoring React components, Next.js pages, data fetching logic, or bundle configurations.

  4. Use the read-github skill

    main

    The read-github skill allows you to read and search GitHub repository documentation and code via the gitmcp.io MCP service.

    Use this skill when:

    • A user provides a GitHub URL.
    • A user mentions a repository in owner/repo format.
    • A user asks questions like "what does this repo do?", "read the docs for X repo", or wants to search code/docs within a specific repository.

  6. Use reusable prompt templates for agent planning and orchestration

    main
    The prompts/ directory contains reusable prompt templates designed to work with the skills and planning workflows in this repository. You can use these as standalone instructions for coding agents or integrate them into your own custom tooling. These prompts are repo-agnostic and can be customized to match your specific internal standards or output formats.

  7. Next.js Code Quality Enforcer capabilities

    main

    The nextjs-code-quality-enforcer provides automated code reviews and real-time feedback to ensure adherence to Next.js standards. It specifically validates:

    • Next.js App Router conventions: Ensures proper use of the modern routing structure.
    • Component Patterns: Validates the distinction and proper usage of Server vs. Client components.
    • File Structure: Enforces proper project organization.
    • TypeScript Usage: Checks for correct type implementations and best practices.
    • Imports: Validates proper import patterns.

  8. Use the OpenAI Docs MCP Skill

    main

    The openai-docs-skill allows you to query the official OpenAI developer documentation via the OpenAI Docs MCP server using a CLI wrapper. Use this skill whenever you need up-to-date, authoritative guidance on OpenAI APIs (Responses, Chat Completions, Realtime, etc.), SDKs, ChatGPT Apps, Codex, MCP integrations, endpoint schemas, parameters, limits, or migrations.

    Core Workflow:

    1. Discover: Use search with a focused query or list to browse the index.
    2. Read: Use fetch on the most relevant URL (optionally including an anchor) to get the content.
    3. Apply: Use the retrieved information to complete your task, ensuring you surface the doc URL used for clarity.
    # Search for a topic
scripts/openai-docs-mcp.sh search "Responses API" 5

# Fetch specific documentation content
scripts/openai-docs-mcp.sh fetch https://platform.openai.com/docs/guides/migrate-to-responses

  9. Use the post-tool/format-javascript-files hook

    main

    The post-tool/format-javascript-files hook automatically formats JavaScript and TypeScript files using Prettier immediately after any Codex Edit operation. It targets .js, .ts, .jsx, and .tsx files modified by Codex.

    Key features:

    • Runs npx prettier --write on modified files.
    • Does not require Prettier to be globally installed (uses npx).
    • Includes error suppression to prevent the hook from failing if Prettier is unavailable.

  10. Understand the Vercel React Best Practices sections

    main

    The vercel-react-best-practices skill is organized into eight distinct sections, each targeting specific performance domains. Rules are grouped by their section ID (the filename prefix). Use these sections to categorize and prioritize performance improvements in your React applications deployed on Vercel.

    Section IDNameImpactDescription
    asyncEliminating WaterfallsCRITICALEliminating sequential awaits to reduce network latency.
    bundleBundle Size OptimizationCRITICALReducing initial bundle size to improve TTI and LCP.
    serverServer-Side PerformanceHIGHOptimizing SSR and data fetching to reduce response times.
    clientClient-Side Data FetchingMEDIUM-HIGHUsing deduplication and efficient patterns to reduce requests.
    rerenderRe-render OptimizationMEDIUMMinimizing unnecessary re-renders to improve responsiveness.
    renderingRendering PerformanceMEDIUMReducing browser workload during the rendering process.
    jsJavaScript PerformanceLOW-MEDIUMMicro-optimizations for hot code paths.
    advancedAdvanced PatternsLOWSpecialized patterns for specific use cases.

  11. Use the Role Creator skill to author Codex custom agents

    main

    The role-creator skill allows you to author, update, or troubleshoot custom Codex agents by managing standalone TOML files. Each agent is defined by a single file which serves as the source of truth for that role.

    Agent Scopes

    • Global: Files are stored in ~/.codex/agents/<agent-name>.toml.
    • Project: Files are stored in <project>/.codex/agents/<agent-name>.toml.

    Note that ~/.codex/config.toml is reserved for global/runtime settings (like thread limits) and should not be used for per-role registration.