Workflows

Provides a structured MCP server to run multi-step workflows with tools, cognitive actions, and state tracking.

javascript

29

GitHub Stars

javascript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "fiveohhwon-workflows-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@fiveohhwon/workflows-mcp"
      ]
    }
  }
}

You can run and interact with a dedicated MCP server that enables LLM-driven workflows. This server lets you define, execute, and monitor multi-step workflows with built-in cognitive actions, tool integrations, and safe, stateful execution. It supports step-by-step execution, dependency-based variable visibility, and efficient payload handling so complex tasks stay predictable and performant.

How to use

To use the MCP server, you run it as a local process or connect to a remote instance via an HTTP endpoint. You then interact with it through an MCP client, which starts workflows, advances steps, and retrieves results. Start a workflow, then feed the output of each step back to the server to continue through the defined sequence. You can observe execution progress, inspect intermediate state, and adjust inputs for iterative runs. The system handles branching, loops, and parallel steps, while keeping strict control over which variables are visible to each step to reduce token usage.

Workflow execution flow

Start a workflow to initialize a session and obtain instructions for the first step. 2. Perform the first step as directed, then provide the result back to the server. 3. Continue with subsequent steps by sending the latest step outcome and requesting the next instruction. 4. Repeat until the workflow completes and a final artifact is produced.

Tips for effective usage

Define clear dependencies to minimize variable visibility and reduce token load.
Leverage differential state updates to focus on changed data.
Use progressive step loading to keep context manageable for long workflows.
Organize steps with meaningful names and explicit inputs/outputs to simplify maintenance.

Available tools

create_workflow

Create a new workflow by providing metadata and initial steps. Returns the created workflow with a unique identifier and version.

list_workflows

List all workflows with optional filtering, sorting, and pagination to help you discover existing automations.

get_workflow

Retrieve a specific workflow by its unique identifier, including its steps and metadata.

update_workflow

Modify an existing workflow, optionally incrementing its version and updating details like description or steps.

delete_workflow

Soft delete a workflow to allow recovery later, preserving historical data.

start_workflow

Begin a workflow execution session and obtain the first-step instructions along with an execution identifier.

run_workflow_step

Execute the next workflow step using the prior step result and continue the execution sequence.

get_workflow_versions

List all saved versions for a workflow to review history and enable rollback.

rollback_workflow

Restore a previous version of a workflow as the active definition and continue execution from that state.