Puppetflow MCP server
{
"mcpServers": {
"puppetflow": {
"type": "http",
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}{
"mcpServers": {
"puppetflow": {
"type": "http",
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}{
"mcpServers": {
"puppetflow": {
"type": "http",
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}{
"servers": {
"puppetflow": {
"type": "http",
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}{
"mcpServers": {
"puppetflow": {
"type": "http",
"url": "https://{instance_host}/api/workspaces/{workspace}/mcp-server/http"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 1.0.0
What it does
Puppetflow is a self-hosted browser automation platform that executes JavaScript workflows and exposes control through an MCP server. Developers use it to build, run, monitor, and manage browser automations, with the ability to watch sessions live and intervene when needed.
- Trigger and monitor Puppetflow browser automation workflows from AI clients with optional JSON input.
- Retrieve run results, logs, screenshots, files, and session recordings from completed workflow executions.
- List and track the status of active and historical browser automation runs.
Tools this server exposes
| Tool | Does |
|---|---|
| trigger_flow | Start a Puppetflow workflow with optional JSON parameters. |
| list_runs | Retrieve a list of workflow runs with their status and details. |
| get_run_result | Fetch the result and logs from a completed workflow run. |
| download_screenshot | Retrieve a screenshot from a workflow run. |
| download_file | Download files generated during a workflow run. |
| download_recording | Get the session recording from a workflow run. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.