exo:hooks

Intercept Exo agent lifecycle events and runtime mutations via HookPoint registrations.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Exo developers need a reliable mechanism to observe, influence, and mutate agent executions without embedding logic in core components.

Core Features & Use Cases

  • HookPoint-based interception across START, PRE/POST_LLM_CALL, PRE_TOOL_CALL, FINISHED, ERROR, and CONTEXT_WINDOW.
  • Runtime mutation of tools, handoffs, and MCP servers, plus message injection for live steering.
  • Use cases include observability, safety rails, dynamic tool loading, and ephemeral context injection for targeted steps.

Quick Start

Register a simple START hook to log the event and demonstrate adding a tool during a run.

Frequently Asked Questions about exo:hooks

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

FAQPage Schema
How do I add observability to agent lifecycles without changing core components?

You can add observability to agent lifecycles by registering HookPoint interceptions across events like START, PRE_LLM_CALL, and FINISHED to monitor runs externally without modifying core logic.

Can I dynamically mutate tools and inject messages during a live agent run?

Yes, you can dynamically mutate tools, handoffs, and MCP servers during a live agent run by using runtime mutation methods and ephemeral message injection via async hooks.

What lifecycle events are available for interception when controlling agent runs?

Available lifecycle events for interception include START, PRE and POST LLM_CALL, PRE_TOOL_CALL, FINISHED, ERROR, and CONTEXT_WINDOW, enabling precise control across execution stages.

How do I register a hook to log agent lifecycle events and add a tool at runtime?

Register a simple START hook via the HookManager API to log the event and demonstrate adding a tool during a run, applying async hook safety semantics for reliable execution.

When should I use hookpoints to build safety rails for agent executions?

Use hookpoints to build safety rails when you need to intercept PRE_TOOL_CALL or ERROR events, allowing you to block unsafe actions or inject corrective context dynamically during execution.

Do I need async hooks to handle runtime mutations across agent runs?

Yes, runtime mutations across agent runs require async hooks to ensure safety semantics and non-blocking execution when intercepting lifecycle events or mutating tools dynamically.