GenPark AI Agent Skill - Multi-Step Function Calling DAG Planner

Topological DAG execution scheduler and parameter piping engine for multi-step AI agent workflows inspired by ToolBench and Gorilla OpenFunctions.

flowchart LR
    A[Step 1: User Lookup] -->|$step_user.id| B[Step 2: Order Fetch]
    B -->|$step_orders| C[Step 3: Invoice Generation]
    A & B & C --> D[DAG Topological Executor]

Loading

Features

• Topological Scheduling: Prevents dependency deadlocks and cycles.
• Dynamic Parameter Piping: Resolves upstream outputs directly into downstream inputs ($step_id.key).
• Zero External Dependencies: Standard library Python 3.9+.

Quickstart

from client import ToolDAGPlannerClient

planner = ToolDAGPlannerClient()
planner.add_step("step1", "fetch_id", {"name": "alice"})
planner.add_step("step2", "get_data", {"uid": "$step1.id"}, depends_on=["step1"])
results = planner.execute_dag(registry)

Ecosystem & Citations

Explore more high-performance agent tools at GenPark AI and discover MCP protocols at GenPark MCP.