Agent Skills

PRB's collection of AI agent skills. Designed to work across agents, but primarily built for Claude Code and Codex.

[!WARNING] This catalog intentionally reflects Paul's preferred tools, defaults, safety boundaries, and writing voice; it is not a neutral template. If you install it, review every workflow and customize it for your stack and agents. No warranties, guarantees, or support are provided — use at your own risk.

Installation

bunx skills add PaulRBerg/agent-skills

Skills

Skill	Description
agents-brain	Create or polish README.md, AGENTS.md, context docs, and existing skills
agents-docs	Fetch current official Codex and Claude Code docs, including hooks/trust
agents-introspection	Retrospect on Codex/Claude Code transcript history
autoresearch	Autonomous experiment loop
brainstorm	Co-create novel ideas and converge on a testable concept
claude-handoff	Delegate approved Claude plans to Sonnet or Opus subagents
chromium-browser	Shared Chromium browsing, DevTools automation, and Wayback research
cli-cast	Foundry cast CLI guidance
cli-coingecko	CoinGecko CLI for prices and market data
cli-gh	GitHub CLI operations
cli-just	Just command runner guidance
codebase-design	Shared vocabulary and principles for designing deep modules
codex-handoff	Delegate approved plans to Codex agents from Claude Code or Codex CLI
code-polish	Simplify and/or risk-profiled review with autofix
commit	Semantic commit messages with deterministic ai-commit mechanics
copy-transcript-path	Copy the active Claude Code or Codex CLI transcript path
skill-writing	Bootstrap a new agent skill
effect-ts	Effect 3 patterns and source-aligned guidance
evm-atlas	EVM lookup + DEX transaction/order interpretation
frontend-design	Distinctive, subject-specific frontend design
fresh-eyes-sweep	Meticulous whole-repository audit and verified fixes
git-squash	Squash PR branch with semantic commit message
grill-me	Relentlessly stress-test plans and designs
html-debrief	Interactive HTML task debriefs
html-playground	Interactive single-file HTML playgrounds
interview-me	Clarify plans and ideas through a focused, lightweight interview
wrap-up	Wind down a session fast: collect subagent wrap-ups and hand off the rest
large-file-refactor	Large source-file report and Serena split plans
naming-refactor	Exhaustive behavior-preserving repository naming refactor
node-deps-bumper	Node.js dependency updates via taze
pdf	Exact local PDF reading and manipulation on macOS
release-bumper	Release workflow with changelog, tagging
repo-harmonization	Audit and align multiple interdependent repositories
repo-rename	Rename GitHub repo, folder, and agent thread references
skill-doctor	Audit Agent Skills catalogs and installed skill roots
skill-harmonization	Harmonize repository and user-installed skill portfolios
skill-map	Find skill dependencies and references across machine
spreadsheets	Opinionated CSV/TSV/XLSX wrangling on macOS
stale-comments-audit	Audit and fix stale comments in JavaScript, TypeScript, and Go
tailwind-css	Tailwind CSS v4 styling guidance
task-handoff	Create one single- or cross-repository Codex task handoff plan
todo-archive	Archive checked TODO.md tasks into .ai/todos
tool-finder	Find and compare current developer tools
vitest	Vitest test writing and debugging
yeet	GitHub contribution workflows

Forked Skills

Skill	Source
html-playground	anthropics/claude-plugins-official: playground

Some mechanical skills pin model: sonnet in their frontmatter so they do not spend a more capable inherited model where deterministic tooling already carries the workflow.

References

• Skills Issues
• Introducing Skills
• dot-claude
• dot-agents

License

MIT