setup-mcps
Configure MCP servers for n8n development. Use when the user says /setup-mcps or asks to set up MCP servers for n8n.
Load before calling the apps tool. Use when the user wants to build or change an n8n App: a small web app, form, dashboard, tracker or other browser UI that n8n hosts. Also load when the conversation lists app targets or a pending app.
Claude Code
$ npx -y skills add n8n-io/n8n --skill app-builder --agent claude-code
Codex
$ npx skills add n8n-io/n8n --skill app-builder -a codex -y
Cursor
$ npx skills add n8n-io/n8n --skill app-builder -a cursor -y
Gemini CLI
$ npx skills add n8n-io/n8n --skill app-builder -a gemini-cli -y
GitHub Copilot
$ npx skills add n8n-io/n8n --skill app-builder -a github-copilot -y
OpenCode
$ npx skills add n8n-io/n8n --skill app-builder -a opencode -y
Windsurf
$ npx skills add n8n-io/n8n --skill app-builder -a windsurf -y
name: app-builder description: >- Load before calling the apps tool. Use when the user wants to build or change an n8n App: a small web app, form, dashboard, tracker or other browser UI that n8n hosts. Also load when the conversation lists app targets or a pending app. recommended_tools: - apps
An n8n App is a static browser app (Vue 3 and Vite). n8n stores its source, builds it in a sandbox and hosts the build on the apps domain. The `apps` tool puts the source of an app into your sandbox workspace and saves it back. You edit the files with the workspace tools.
agents yet. Keep data in component state or `localStorage`.
`@vitejs/plugin-vue`. You cannot change `package.json` or `package-lock.json`. Write routing, state and UI components yourself.
and the published app run on different hosts.
A new app starts from this source:
Add components under `src/`. Keep files small and focused.
1. **Create or open the app.**
subdomain label of the app URL: lowercase letters, digits and single hyphens. Derive it from the name.
reuses the pending id.
conversation has no id, call `list` (optionally with a `search` text) and pick the app by name. If several apps match, ask the user which one. Then call `open`.
folder that is already open, with your uncommitted edits. 2. **Edit the folder in batches.** Read with `workspace_read_files` and change files with `workspace_edit_files`. Put all the files of one step into one call, not one call for each file. Use the single-file tools only for a single file. Read only what you need: use `grep` to find text, and read line ranges of large files. `workspace_edit_files` runs every operation and reports each one, so fix only the operations that failed. Never change `package.json` or `package-lock.json`. 3. **Build before you commit.** In the app folder, run `npm ci` once, then `npm run build`. Fix every error and build again until the build passes. Use `npm ci`, not `npm install`: `npm install` can change `package-lock.json`, and n8n then rejects the commit. 4. **Commit once for each finished change.** A commit saves the folder (without `node_modules` and `dist`) as one version, and n8n builds it again. Never commit half-done work: every commit is a version. 5. **Check the build.** A commit returns `buildStatus`. On `failed`, read `diagnostics` (stage, message, log), fix the source and commit again. After three failed builds in a row, stop and explain the problem to the user.
If a commit reports that the app draft changed, someone else saved a newer version. Your folder is kept. Note your changes, call `discard` to load the newer draft, and apply your changes again. `discard` keeps `node_modules`, so you do not need to run `npm ci` again.
One conversation can build several apps for one requirement, for example a public dashboard and an admin app. Create one app for each audience or access level. Each app has its own folder. Commit each app separately and always pass the right `appId`.
You cannot publish apps. Tell the user to publish with the Publish button of the app. A new app is visible to the members of its project. The user can change this in the app settings.
Fair-code platform to build and deploy AI agents and workflows. Combine a visual canvas with custom code, run it self-hosted or in the cloud, and connect to 1500+ integrations. AI automation you can trust with real work, from prototype to production.
Configure MCP servers for n8n development. Use when the user says /setup-mcps or asks to set up MCP servers for n8n.
Use the n8n CLI to manage workflows, credentials, executions, and more on an n8n instance. Use when the user asks to interact…
Load immediately after an Agent intent, before the first agent_builder_* call. Governs how you create, configure, test, and…
Load before calling the data table tools or parse-file. Use for natural standalone requests like "what data tables do I have?",…
Debug failed or wrong-output workflow executions using executions tools. Load when the user reports execution failures,…
Load when the request depends on what is already on this instance rather than on what the user just typed: a short or ambiguous…
Direct browser control via CDP for web interaction: automation, scraping, testing, screenshots, and site/app work. Also covers…
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling…
Add local office-document-to-Markdown conversion to NanoClaw agent containers with the pinned Firecrawl AnyDoc CLI. Use when…
Build or improve an OpenCLI JS adapter using observed site behavior, existing adapters and browser API evidence. For ad-hoc…