Skillget

awesome-agent-skill

mcp
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 6 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Curated skill pack for LLM agents in engineer and science workflow (Cursor & Claude ready).

README.md
awesome-agent-skill — reusable skills for agentic developer tools

Stars
License: MIT
Skills
Formats

A curated, open-source skill library for agentic developer tools.

Give your coding agent reusable playbooks for planning, debugging, QA, security, docs, data work, browser automation, and shipping.

Why This Exists

Most agents start every task from a blank prompt. Skills give them reusable operating procedures: when to ask for clarification, how to run a review, how to triage tests, how to handle PDFs, how to build a quick analysis, how to use browser QA, and how to ship safely.

awesome-agent-skill packages those workflows as portable SKILL.md files for modern coding agents.

What You Get

Area Includes
Agent workflow Specs, planning, TDD, incremental implementation, code review, verification
Agent systems MCP, RAG, tool contracts, context-window management, agent evaluation
Browser QA gstack-style browsing, QA, canary checks, benchmarks, screenshots
Reliability and security CI gates, launch checklists, observability, API security, skill supply-chain audit
Data and content Analysis, visualization, documents, spreadsheets, PDFs, presentations
Role playbooks Engineering, research, product, operations, orchestration, language experts

Quickstart

Clone the pack:

git clone https://github.com/charlieviettq/awesome-agent-skill.git
cd awesome-agent-skill

Install a few Cursor skills into a project:

mkdir -p /path/to/project/.cursor/skills
rsync -a .cursor/skills/core-workflow/ /path/to/project/.cursor/skills/core-workflow/
rsync -a .cursor/skills/security-appsec/ /path/to/project/.cursor/skills/security-appsec/
rsync -a .cursor/skills/writing-docs/ /path/to/project/.cursor/skills/writing-docs/

Install Claude Code skills:

mkdir -p /path/to/project/.claude/skills
rsync -a .claude/skills/ /path/to/project/.claude/skills/

Reload your agent session after copying skills.

Formats

Agent surface Path Status
Cursor .cursor/skills/**/SKILL.md Source of truth
Claude Code .claude/skills/**/SKILL.md Generated and committed

Regenerate Claude-format skills after editing Cursor-format skills:

python3 scripts/convert-to-claude.py --in-repo --force --write-map

Skill Map

Skill map for awesome-agent-skill

Full index: SKILL_INVENTORY.md

Repository Layout

.
├── .cursor/skills/      # Cursor skill format, source of truth
├── .claude/skills/      # Claude Code skill format, generated from Cursor skills
├── scripts/             # Conversion and maintenance scripts
├── .github/workflows/   # Lightweight validation
└── SKILL_INVENTORY.md   # Human-readable skill index

Highlights

Folder Good For
core-workflow/ Spec-first implementation, planning, TDD, verification, reviews
ai-agent-systems/ MCP servers, RAG systems, agent evals, tool schemas
gstack/ Browser QA, ship workflows, design review, scrape flows
voltagent/ Role-based subagent playbooks
security-appsec/ API security, secure design, skill supply-chain checks
reliability-ops/ CI gates, SLOs, launch readiness, postmortems
writing-docs/ PDF, DOCX, XLSX, PPTX, prose polish
visualization/ Charts, figures, infographics, data storytelling

Design Principles

  • Portable: skills are plain folders with SKILL.md.
  • Composable: copy one domain or the full pack.
  • Public-safe: no secrets, customer data, or private org assumptions.
  • Agent-first: written as workflows an agent can follow, not as static articles.
  • Reviewable: skills stay small; long references belong in reference.md or examples.

Updating Skills

Edit the Cursor source:

$EDITOR .cursor/skills/core-workflow/spec-driven-development/SKILL.md

Regenerate Claude output:

python3 scripts/convert-to-claude.py --in-repo --force --write-map

Run the lightweight check:

python3 - <<'PY'
from pathlib import Path
skill_files = list(Path(".cursor/skills").rglob("SKILL.md")) + list(Path(".claude/skills").rglob("SKILL.md"))
if not skill_files:
    raise SystemExit("No skills found")
bad = [path for path in skill_files if not path.read_text(encoding="utf-8").startswith("---")]
if bad:
    raise SystemExit("\\n".join(map(str, bad)))
print(f"Validated {len(skill_files)} skill files")
PY

Private Skills

Keep team-specific or sensitive skills in your own repository under paths such as .cursor/skills/private/ and .claude/skills/private/. This public pack is intentionally generic.

Contributing

Contributions are welcome. A good skill is:

  • Narrow enough for an agent to know when to use it.
  • Explicit about triggers, workflow, inputs, outputs, and verification.
  • Free of secrets, private URLs, customer data, and organization-specific assumptions.
  • Useful without requiring hidden context.

Open a PR with a short explanation of the workflow the skill improves.

License

MIT. Use it, fork it, and adapt it for your own agents.

Reviews (0)

No results found