Skip to content

MCP Server

Dagu includes a built-in Model Context Protocol (MCP) server. There is no separate Dagu MCP package to install: run the Dagu HTTP server and point an MCP client at the /mcp endpoint.

Use MCP when an AI tool should operate a running Dagu server: inspect workflows, maintain Markdown documents, read run state, preview changes, apply DAG or document edits, or control DAG runs through the same authenticated server boundary as the REST API.

What Dagu Exposes

The MCP server is intentionally compact:

SurfacePurpose
Toolsdagu_read, dagu_change, and dagu_execute cover reading state, maintaining documents and DAG YAML, and controlling runs.
Resourcesdagu://... resources expose DAG specs, Markdown documents, run details, run logs, and built-in MCP references.
PromptsBuilt-in prompts guide DAG and document authoring, editing, and failed-run debugging workflows.
Audit eventsMCP requests, tool calls, subscriptions, and downstream DAG actions are audit-attributed to the accepted credential.

The server uses Streamable HTTP. The default local URL is:

text
http://localhost:8080/mcp

If the Dagu server uses a base path, place /mcp under that base path. For example, a server mounted at /dagu exposes MCP at /dagu/mcp.

Common Flows

GoalStart here
Connect a client for the first timeQuickstart
Configure Claude Code, Codex, Cursor, VS Code, or another clientClients
Understand the request path and audit flowArchitecture
Choose API key, role, surface, and attribution settingsAuthentication
Read or mutate Dagu through MCP toolsTools
Use dagu:// resources or run-completion subscriptionsResources
Review what appears in audit logsAuditability

Skill Vs MCP

The Dagu skill and the Dagu MCP server solve related but different problems.

IntegrationConfigureBest for
Dagu skillgh skill install dagucloud/dagu daguTeaching AI coding tools how to write valid Dagu workflow YAML.
Dagu MCP serverhttp://localhost:8080/mcpLetting MCP clients read Dagu state and documents, apply scoped edits, and control runs.

Most AI-assisted workflow authoring setups benefit from both: install the skill for authoring guidance, then connect MCP when the tool should operate a running Dagu server.

Dagu is open source under the GNU General Public License v3.0.