GitHub user FenjuFu created a discussion: Community MCP server for DolphinScheduler: iflytek/dolphin-mcp-pilot
Sharing a community integration for anyone running DolphinScheduler with LLM agents. [iflytek/dolphin-mcp-pilot](https://github.com/iflytek/dolphin-mcp-pilot) is an MCP (Model Context Protocol) server that lets an AI client drive DolphinScheduler through its Open API. Rather than read-only wrappers, it exposes ~53 tools covering real operations: - Projects, workflow and DAG creation (SQL and DAG-based) - Schedule lifecycle (cron create / online / offline) - Instance control 鈥?pause, resume, rerun, failure recovery - Resource content management with version rollback - Task-level logs and monitoring - Raw API passthrough as a fallback for anything not yet wrapped Stack: Python + FastMCP, Docker image provided. Auth via DolphinScheduler API token or user/password, with per-request credentials in HTTP mode for multi-tenant setups. It is community-maintained and not an official Apache DolphinScheduler component 鈥?check its compatibility and security notes before pointing it at production, and use least-privilege credentials for agent access. Related: I've proposed documenting community MCP integrations under `docs/integration` in #18551 (PR #18552), covering this project alongside ocean-zhc/dolphinscheduler-mcp. Feedback on either is welcome. GitHub link: https://github.com/apache/dolphinscheduler/discussions/18555 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected]
