pipefy-toolkit-setup — Claude Code skill: how to install
Guides you through the initial Pipefy AI toolkit setup across Claude Code or Cursor. It helps pick the right integration path and finishes authentication.
npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a claude-codeWhat pipefy-toolkit-setup does
The skill acts as an onboarding guide for connecting your environment to Pipefy. You decide the architecture. Instead of guessing, the assistant asks for your preferred integration path first. Options include the hosted MCP at mcp.pipefy.com without local Python, the Cursor Marketplace plugin, a local install.sh setup, or a CLI-only binary.
Once you confirm a path, the agent carries out the official installation steps without inventing alternate commands. It guides you through authentication and checks for conflicting server registrations. A read-only list_organizations call verifies that your credentials work before you start modifying data.
Do not call this skill for everyday Pipefy operations. Once your MCP server runs, switch to domain-specific skills for handling cards, pipes, or tables. You also need an existing Pipefy account to complete setup, including admin permissions if you require a service account.
What it helps with
- choose between hosted MCP, client plugins, and local CLI
- connect Claude Code or Cursor to
mcp.pipefy.comwithout Python - install the local Pipefy toolkit using
install.sh - confirm access rights with
list_organizations
Example prompts for the agent
- Help me set up the Pipefy MCP connection
- Connect Claude Code to mcp.pipefy.com
- Run initial onboarding for Pipefy AI toolkit
How to install pipefy-toolkit-setup
1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a claude-code2Manually: put the files in place
~/.claude/skills/pipefy-toolkit-setup/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/pipefy-toolkit-setup/
1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a codex2Manually: put the files in place
.agents/skills/pipefy-toolkit-setup/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a cursor2Manually: put the files in place
.cursor/skills/pipefy-toolkit-setup/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install pipefy/ai-toolkit pipefy-toolkit-setup2Manually: put the files in place
.github/skills/pipefy-toolkit-setup/SKILL.md1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a gemini-cli2Platform CLI
gemini skills install https://github.com/pipefy/ai-toolkit --path plugins/pipefy/skills/pipefy-toolkit-setupManually: put the files in place
.devin/skills/pipefy-toolkit-setup/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about pipefy-toolkit-setup
How do I install pipefy-toolkit-setup in Claude Code?
Run npx skills add pipefy/ai-toolkit --skill pipefy-toolkit-setup -a claude-code in your terminal. This creates the skill definition at ~/.claude/skills/pipefy-toolkit-setup/SKILL.md.
What does pipefy-toolkit-setup do?
It walks you through configuring the Pipefy AI toolkit step by step. The assistant prompts for your preferred setup method, executes the canonical commands, and verifies your credentials.
Should I keep using this skill after MCP is configured?
No, this skill handles only first-time setup and onboarding. Once your connection works, use dedicated domain skills for pipes and database tables.
Source text
SKILL.md · Apache-2.0
--- name: pipefy-toolkit-setup description: > Guide a first-time user through Pipefy AI toolkit setup (Cursor Marketplace plugin, hosted MCP, local install.sh, or Claude Code plugin). Use when the user asks to install Pipefy MCP, connect Claude/Cursor to Pipefy, run /pipefy:install, or set up mcp.pipefy.com. Do not use for day-to-day Pipefy workflows after MCP is already working. tags: [pipefy, onboarding, install, mcp, setup, claude, cursor] metadata: surfaces: "mcp cli" --- # Pipefy toolkit setup (first-time onboarding) Choose a connection path below, then read the [MCP reference](references/mcp.md) or [CLI reference](references/cli.md) for that path. Load only the relevant reference. **Canonical install snippets** live in [`docs/install.md`](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md). The root README has a quick-start table with the first command for each install path. This skill is the agent **checklist**: print or run the install-guide blocks verbatim, and do not invent alternate commands. Edge cases: [`packages/mcp/README.md`](https://github.com/pipefy/ai-toolkit/blob/main/packages/mcp/README.md). Auth: [`docs/cli/auth.md`](https://github.com/pipefy/ai-toolkit/blob/main/docs/cli/auth.md). Env: [`docs/config.md`](https://github.com/pipefy/ai-toolkit/blob/main/docs/config.md). ## When to use - “Set up Pipefy”, “install the MCP”, or “connect Claude/Cursor to Pipefy”. - The user mentions a client plugin, a hosted connection, or a fresh machine without toolkit access. **Do not use** once tools already work — switch to a domain skill (pipes, tables, …). ## Prerequisites - User has a Pipefy account (Admin access if they need a service account). ## Steps 1. **Choose one path** — ask the user; do not pick silently. Respect whether they want a hosted connection, a client plugin, a local toolkit, or CLI only. If they only say “Claude Code”, ask: Hosted MCP (zero local Python) or the Claude Code plugin (slash commands and local CLI)? If unsure, suggest Hosted. If they only say “Cursor”, ask: Cursor Marketplace plugin (hosted, no local Python) or Quick install (local CLI and tools that read local files)? If unsure, suggest the plugin. | Path | Install guide section | Outcome | |------|----------------|---------| | Cursor Marketplace plugin | [Cursor Marketplace plugin](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md#6-cursor-marketplace-plugin) | Hosted server, browser sign-in, no local Python | | Hosted MCP | [Hosted MCP](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md#1-hosted-mcp-claude-code) | HTTPS `mcp.pipefy.com` | | Local toolkit | [Quick install](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md#3-quick-install-script) | Local server and CLI, including local-file tools | | Claude Code plugin | [Claude Code plugin](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md#2-claude-code-plugin) | Marketplace and slash install/login | | CLI only | [CLI](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md#4-cli-only) | `pipefy` on PATH; no MCP | 2. **Execute the canonical install instructions** for that path. Keep client setup and local installation aligned with the user's choice. 3. **Authenticate** using the selected path's login flow. 4. **Verify** access with a permitted read-only organization operation (`list_organizations` needs no id and surfaces organization ids), or the local installation check. Check for conflicting registrations using the selected reference. ## Success criteria - Exactly the selected connection is configured, with no conflicting registrations; CLI-only needs no server registration. - Authentication completed. - The permitted read-only operation or local installation check succeeds. ## See also - [`docs/install.md`](https://github.com/pipefy/ai-toolkit/blob/main/docs/install.md) - [`docs/uninstall.md`](https://github.com/pipefy/ai-toolkit/blob/main/docs/uninstall.md) — `uninstall.sh --scan`, teardown, and switching between hosted, local, and plugin - [`plugins/pipefy/README.md`](https://github.com/pipefy/ai-toolkit/blob/main/plugins/pipefy/README.md)
© pipefy · Apache-2.0 · pipefy/ai-toolkit