Versioned, runnable lists — checklists, procedures, courses and agent skills. Fork one, improve it, run it step by step.
adduser deployfnm install --ltsnpm ci && npm run buildcaddy reverse-proxy --to :3000Every list is a living standard: start from the best one, adapt it, and follow it step by step.
Search by keyword or meaning across the public library.
Fork it, adapt it, and suggest your change back — the list gets better for everyone.
Follow it step by step, check things off, and keep a history of your runs.
Any public list exports as a SKILL.md package in the open Agent Skills format — the one Claude Code, Codex, Cursor, Gemini CLI and dozens of other agents read. One command installs it.
$ npx skills add https://setfork.com/<handle>/<slug>/skill.tar.gz <slug>/ ├── SKILL.md name, description, steps ├── scripts/ author scripts ├── references/ docs loaded on demand └── assets/ templates, data
When nothing matches your search, AI experts draft candidate lists. You pick one, refine it and publish a version — nothing goes public without a person.
Every change is a version backed by git. Suggest edits like pull requests, comment on single steps, and keep credit honest.
Connect over MCP: agents search lists, patch single steps, publish skills and record runs. Listed in the official MCP registry.
{
"mcpServers": {
"setfork": {
"type": "http",
"url": "https://setfork.com/api/mcp",
"headers": {
"Authorization": "Bearer sf_…"
}
}
}
}An email when the lists you saved get better.
Public content is checked before it goes live.
Everything AI writes starts as a draft.

Free to browse, fork and run. Bring your agent along over MCP.