YAML Workflow — Claude Code MCP server: what it does, how to install
YAML Workflow is a Claude Code MCP server that turns declarative YAML pipelines into agent tools. Runs build steps and data workflows directly over stdio.
MIT ★ 5 updated 2026-10-12
claude mcp add yaml-workflow -- uvx yaml-workflowWhat YAML Workflow does
YAML Workflow exposes local task pipelines to the AI model as executable functions. The engine handles various task types, including shell scripts, inline Python code, HTTP calls, Jinja2 templates, and batch operations. Execution state is tracked in local files, allowing failed runs to resume without restarting from scratch.
Workflows support parallel execution via depends_on, step composition with imports, and dry-run previews to inspect commands beforehand. Running the server requires Python and the uvx runner in your environment.
What it helps with
- run multi-step build workflows directly from the model interface
- orchestrate shell tasks and Python code inside versioned YAML files
- preview workflow execution steps without modifying files using dry-run
- resume interrupted batch pipelines from the last successful step
Example prompts for the agent
- Run the workflow at
workflows/hello_world.yamlwith parametername=Alice - Execute a dry-run preview for
workflows/data_pipeline.yaml - Check the execution status of the current YAML pipeline
How to install YAML Workflow
Platform CLI
claude mcp add yaml-workflow -- uvx yaml-workflowuvx runs the command named after the package; if the README names another one, use uvx --from yaml-workflow <command>
Platform CLI
codex mcp add yaml-workflow -- uvx yaml-workflowuvx runs the command named after the package; if the README names another one, use uvx --from yaml-workflow <command>
Config file
{
"mcpServers": {
"yaml-workflow": {
"command": "uvx",
"args": [
"yaml-workflow"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from yaml-workflow <command>
Config file
{
"servers": {
"yaml-workflow": {
"type": "stdio",
"command": "uvx",
"args": [
"yaml-workflow"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from yaml-workflow <command>
Platform CLI
gemini mcp add yaml-workflow uvx yaml-workflowuvx runs the command named after the package; if the README names another one, use uvx --from yaml-workflow <command>
Step 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 YAML Workflow
How to install YAML Workflow in Claude Code?
Run claude mcp add yaml-workflow -- uvx yaml-workflow in your terminal. This registers the server over stdio using uvx.
What does YAML Workflow do?
It translates YAML-defined pipeline steps into callable tools for the agent. The model can run tasks, inspect dependencies, and pass parameters.
What is required to run YAML Workflow?
You need Python, the uvx tool, and the PyYAML and Jinja2 packages installed in your environment.