ax-agent

Configure sandboxed AI agent processes with IPC and tool management.

5|2|Updated Feb 8, 2026
One-click install
npx skills add https://github.com/project-ax/ax-legacy --skill ax-agent-project-ax
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ax-agent
Source: https://github.com/project-ax/ax-legacy/tree/main/.claude/skills/ax-agent
Command: npx skills add https://github.com/project-ax/ax-legacy --skill ax-agent-project-ax

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

The ax-agent Skill solves the problem of managing a sandboxed agent process by providing tools for runner configuration, IPC client interaction, local and IPC tool handling, prompt building, and identity loading.

Core Features & Use Cases

  • Process Configuration: Modifies and configures the sandboxed agent process for different tasks such as runner dispatch, IPC transport selection, and prompt building.
  • IPC Client Interaction: Enables interaction with IPC clients for tasks that require deterministic processing.
  • Local and IPC Tool Handling: Manages tools for operations within the agent process, including local sandboxes and proxy-based tools.
  • Prompt Building: Builds system prompts from modular components for effective LLM agent loop operations.
  • Identity Loading: Loads identity files for user and skill operations.

Quick Start

Load identity files for the ax-agent process with 'load-identity ax-agent-identity-file.json'.

Frequently Asked Questions about ax-agent

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

FAQPage Schema
How do I configure a sandboxed AI agent process for custom tasks?

Configuring a sandboxed AI agent process involves modifying its operational settings for runner dispatch, IPC transport selection, and prompt building to customize agent behaviors for specific tasks.

What is IPC client interaction in sandboxed agent processes?

IPC client interaction enables a sandboxed agent process to communicate with IPC clients, ensuring deterministic processing for operational tasks that require reliable inter-process communication.

How do I build system prompts from modular components for LLM agents?

Building system prompts from modular components involves constructing prompts within the agent process configuration to enable effective LLM agent loop operations and customized behaviors.

Can I manage both local and proxy-based tools within an agent process?

Yes, you can manage local and IPC tools within the agent process, handling operations that include local sandboxes and proxy-based tools for customized agent task execution.

How do I load identity files for sandboxed user and skill operations?

Loading identity files requires executing the identity loading command with the specified JSON file, enabling user and skill operations within the sandboxed agent process environment.

Does this approach support runner configuration and dispatch for sandboxed operations?

Yes, the sandboxed agent process modification supports runner configuration and dispatch, allowing you to select IPC transports and configure operational tasks effectively.