Skills Configuration

Skills provide specialized instructions and techniques that agents can follow. AgentPool supports automatic injection of skills into agent system prompts using structured XML formatting.

Overview¶

Skills injection allows you to:

Automatically include relevant skill instructions in agent prompts
Configure global defaults for all agents
Override per-agent using skills tool configuration
Limit the number of skills included to manage token count

Configuration Structure¶

Global Skills Configuration¶

skills:
  # Skill discovery paths
  paths:
    - ~/.config/agentpool/skills
    - ./skills

  # Include default AgentPool skills (default: true)
  include_default: true

  # Instruction injection configuration
  instruction:
    # Full-instruction injection mode: description, matcher, or all
    inject: description
    # Maximum number of skills to inject (default: 20)
    max_skills: 20

Injection Modes¶

The static <available-skills> catalog (skill name + description) is always emitted when the pool loads skills. The inject field controls whether the full skill instructions (<skill_content>) are also placed into the system prompt, which costs more tokens.

Mode	Behavior	Use Case
`description` (default)	Catalog only — NO full `<skill_content>`. Full instructions are loaded on demand via `load_skill`.	Token-saving; the common default
`matcher`	Catalog + full instructions for skills selected by a runtime `matcher_fn`	Agents that should auto-load relevant skills without waiting for a tool call
`all`	Catalog + full instructions for every skill	When skills contain critical instructions that must always be present

Agent-Specific Overrides¶

Per-agent overrides are handled by the (deprecated) type: skills toolset, which is a no-op since skills tools are auto-provided by SkillManagerCap. The inject/max_skills settings are global; there is no per-agent injection_mode override.

XML Output Format¶

When skills injection is enabled, agents receive a skill catalog in their system prompt:

<available-skills>
  <skill name="python-style-guide" description="PEP 8 coding conventions" />
  <skill name="refactoring" description="Safe refactoring techniques" />
</available-skills>

With inject: all (or matcher-selected skills), full content is also injected:

<skill_content name="python-style-guide">
  ## Python Style Guide

  Follow PEP 8 conventions:
  - Use 4 spaces for indentation
  - Maximum line length of 100 characters
</skill_content>

Complete Example¶

# Global skills configuration
skills:
  paths:
    - ~/.config/agentpool/skills
    - ./project-skills
  include_default: true

  # Default: catalog-only injection for all agents (token-saving)
  instruction:
    inject: description
    max_skills: 20

Behavior Notes¶

The <available-skills> catalog is emitted whenever the pool has visible skills — this is not disabled by default.
inject: matcher requires a programmatically-provided matcher_fn (it is not serializable in YAML). Without one it falls back to description and logs a warning.
Full instructions are always available on demand via the load_skill tool, regardless of the inject mode.

Toolsets - Configure agent tools including skills tool
Agent Pool - Global pool configuration