OpenAI Skills - Setup Guide¶
Source: openai/skills (703K+ combined installs) Category: Development / Security / Design Quality Tier: 🟢 Production
Official skills published by OpenAI for use with Claude Code and compatible agents (including Hermes). These are production-grade, well-maintained workflows covering PDF generation, CI debugging, security auditing, project management (Linear), browser testing (Playwright), and design implementation (Figma).
Installation¶
npx skills add openai/skills --skill pdf
npx skills add openai/skills --skill gh-fix-ci
npx skills add openai/skills --skill security-best-practices
npx skills add openai/skills --skill linear
npx skills add openai/skills --skill playwright
npx skills add openai/skills --skill figma-implement-design
Included Skills¶
| Skill | Installs | Purpose |
|---|---|---|
| 10.5K | Read, create, and review PDFs with rendering and layout verification | |
| gh-fix-ci | 7.8K | Debug failing GitHub Actions PR checks using gh CLI |
| security-best-practices | 5.8K | Language/framework-specific security reviews with auto-detection |
| linear | 5.2K | Manage Linear issues, cycles, and projects from the CLI |
| playwright | 4.7K | Browser automation and end-to-end testing with Playwright |
| figma-implement-design | 4.4K | Convert Figma designs into production code |
Key Capabilities¶
pdf - PDF Manipulation¶
- Render PDFs to PNGs for visual inspection (
pdftoppm) - Generate PDFs programmatically with
reportlab - Extract text with
pdfplumberandpypdf - Quality-first: verify alignment, spacing, and legibility before delivery
- Dependencies: poppler-utils, reportlab, pdfplumber, pypdf
gh-fix-ci - CI Debugging¶
- Inspect failing PR checks via
gh pr checks - Fetch GitHub Actions logs for actionable failures
- Bundled
inspect_pr_checks.pyscript handles API field drift - Creates fix plans with approval gates before implementation
- Scope: GitHub Actions only (external providers like Buildkite are out of scope)
security-best-practices - Security Auditing¶
- Auto-detects language and framework from project context
- Loads reference files for Python, JavaScript/TypeScript, and Go
- Three modes: secure-by-default coding, passive vulnerability detection, full security reports
- Produces prioritized reports with severity classifications
- Supports project-specific overrides with documentation
linear - Project Management¶
- Manage Linear issues, cycles, and project tracking
- CLI-driven workflow for ticket creation and status updates
playwright - Browser Testing¶
- End-to-end browser automation and testing
- Cross-browser testing with Playwright's API
figma-implement-design - Design-to-Code¶
- Convert Figma design files into production-ready code
- Handles component extraction, styling, and layout implementation
Quick Start¶
# PDF generation
npx skills use openai/skills@pdf
# Debug failing CI
npx skills use openai/skills@gh-fix-ci
# Security audit
npx skills use openai/skills@security-best-practices
Verification¶
npx skills list | grep "openai/skills"
# Expected: 6 skills listed with install counts
Notes¶
- These are official OpenAI skills - well-maintained and updated regularly
- All skills use
npx skills addfor installation (standard marketplace workflow) - The
security-best-practicesskill is especially valuable for Hermes agents working on production codebases gh-fix-ciincludes a bundled Python script for reliable CI inspection (works aroundghCLI field drift)- PDF skill prefers visual verification - renders to PNG before declaring output ready
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.