froggo-mission-control: Skill for Claude Code

Workflow Automator Skill

Overview

Workflow Studio is the platform's visual automation system. Agents can create, edit, execute, and manage workflows via the REST API at /api/local/workflows.

API Endpoints

List Workflows

GET /api/local/workflows?limit=50&offset=0
→ { workflows: WorkflowSummary[] }

Create Workflow

POST /api/local/workflows
Body: { name: string, description?: string, state: SerializedWorkflow }
→ { id: string, name: string, ... }

Update Workflow

PUT /api/local/workflows/:id
Body: { name?, description?, state?, is_deployed?: 0|1 }
→ WorkflowDetail

Delete Workflow

DELETE /api/local/workflows/:id

Execute Workflow

POST /api/local/workflows/:id/execute
Body: { input?: any }
→ { id: string, workflowId: string, status: string, result?: any }

List Executions

GET /api/local/workflows/:id/executions?limit=50&offset=0
→ { executions: ExecutionResult[] }

SerializedWorkflow Format

Every workflow is a JSON object with this structure:

{
  "version": "1",
  "blocks": [
    {
      "id": "unique-string-id",
      "position": { "x": 100, "y": 200 },
      "config": {
        "tool": "block_type",
        "params": { /* block-specific parameters */ }
      },
      "inputs": {},
      "outputs": { "response": "string" },
      "metadata": {
        "id": "same-as-block-id",
        "name": "Human-readable name",
        "icon": "LucideIconName",
        "color": "#hexcolor",
        "category": "trigger|ai|logic|integration|data|utility"
      },
      "enabled": true
    }
  ],
  "connections": [
    { "source": "block-id", "target": "block-id" }
  ],
  "loops": {}
}

Block Types Reference

Triggers

Type	Params
starter	triggerType ("manual"|"schedule"|"webhook"|"api"), schedule (cron), webhookPath
generic_webhook	webhookPath, method, secret

AI

Type	Params
agent	agentId (MC agent), model, systemPrompt, userPrompt, temperature, maxTokens
thinking	model, prompt, maxTokens
openai	model, systemPrompt, userPrompt, temperature, maxTokens
evaluator	criteria, rubric, model
guardrails	rules, action ("block"|"warn"|"retry")