Puppetflow — Claude Code MCP server: what it does, how to install
Controls remote Puppetflow browser routines directly from AI chat sessions. Helps trigger JavaScript flows, inspect execution logs, and retrieve screenshots.
Open on GitHubWhat Puppetflow does
Puppetflow connects AI agents to a self-hosted browser automation platform. All automation steps execute on your dedicated instance, so your local machine stays completely unburdened. Your agent searches existing JavaScript routines, initiates new runs, and tracks execution status through streamable HTTP endpoints.
The integration exposes tools to inspect run histories, review full console logs, and download captured session artifacts. Whenever a flow requires human intervention, you can watch the session live and take manual control. Without an active self-hosted instance and an issue-specific access token, the client cannot communicate with the API. Configuration works via Bearer token authentication or remote OAuth with PKCE.
What it helps with
- trigger automated browser flows from assistant chat
- retrieve error logs and execution artifacts after a failed run
- download session recordings and captured screenshots
- pass custom parameters into JavaScript workflows
Example prompts for the agent
- Run the onboarding check flow and return the final screenshot
- Inspect the latest workflow run and report any logged failures
- List available browser automations in my workspace and execute the report fetcher
How to install Puppetflow
1Template address
The server address is a template: replace {…} or […] with your own values
2Platform CLI: remote server
claude mcp add --transport http mcp https://{instance_host}/api/workspaces/{workspace}/mcp-server/http1Template address
The server address is a template: replace {…} or […] with your own values
2Platform CLI: remote server
codex mcp add puppetflow --url https://{instance_host}/api/workspaces/{workspace}/mcp-server/http1Template address
The server address is a template: replace {…} or […] with your own values
2Config file
{
"mcpServers": {
"puppetflow": {
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}1Template address
The server address is a template: replace {…} or […] with your own values
2Config file
{
"servers": {
"puppetflow": {
"type": "http",
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}1Template address
The server address is a template: replace {…} or […] with your own values
2Platform CLI: remote server
gemini mcp add -t http puppetflow https://{instance_host}/api/workspaces/{workspace}/mcp-server/httpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Puppetflow
How to install Puppetflow in Claude Code?
Run claude mcp add --transport http puppetflow https://{instance_host}/api/workspaces/{workspace}/mcp-server/http in your terminal. Replace {instance_host} and {workspace} with your actual server host and workspace slug.
What does the Puppetflow MCP server do?
It allows AI clients to interact with a self-hosted Puppetflow instance. The agent can trigger JavaScript automations, check run progress, and fetch logs or recordings.
Where do I get the access token for setup?
Log in to your Puppetflow web interface, navigate to Workspace > Settings > Instance MCP, and generate a new token to authorize your client requests.
