exo:ptc

Batch tool calls inside Exo agents to minimize LLM round-trips.

1|Updated Feb 16, 2026
One-click install
npx skills add https://github.com/midsphere-ai/exo --skill exo-ptc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: exo:ptc
Source: https://github.com/midsphere-ai/exo/tree/main/skills/ptc
Command: npx skills add https://github.com/midsphere-ai/exo --skill exo-ptc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

PTC reduces LLM round-trips by enabling programmatic tool calling inside a restricted Python namespace, batching multiple tool invocations in a single execution.

Core Features & Use Cases

  • PTC enables batching of tool calls via the exo_ptc executor, cutting round-trips and keeping intermediate data out of the model context.
  • Supports HITL-aware flows, swarm propagation, and dynamic tool usage by routing calls through default_api in a sandboxed environment.
  • Use cases include parallel data gathering, batch processing across regions, and workflows that require aggregated results from multiple tools.

Quick Start

Enable PTC on an agent and write Python code that batches tool calls inside exo_ptc to minimize LLM round-trips.

Frequently Asked Questions about exo:ptc

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I batch tool calls for LLM agents to reduce round-trips?

Batch tool calls for LLM agents by executing multiple tool invocations programmatically inside a restricted Python sandbox. This minimizes LLM round-trips and keeps intermediate data out of the model context.

What is programmatic tool calling in a sandboxed agent environment?

Programmatic tool calling in a sandboxed agent environment routes multiple tool invocations through a default API within a restricted Python namespace, enabling dynamic tool usage without direct model interaction for each step.

How do I enable batch processing across multiple regions for AI workflows?

Enable batch processing across multiple regions by configuring an agent runtime executor to gather parallel data and aggregate results from multiple tools in a single execution.

Can I use human-in-the-loop flows with batched agent tool calls?

Yes, batched agent tool calls support human-in-the-loop aware flows, allowing dynamic tool usage and swarm propagation across agents while executing within a restricted sandbox environment.

What are the limitations of batching tool calls inside a restricted Python namespace?

Batching tool calls inside a restricted Python namespace requires a specific runtime executor and default API exposure, limiting execution to the sandboxed environment and preventing access to unexposed external resources.

Does the exo:ptc skill work without the ptc runtime executor?

No, the exo:ptc skill requires the ptc runtime, the dedicated executor, and tools exposed via default API to function properly within the restricted sandbox environment.