Skip to content
SDKs
PY Python SDK TS TypeScript SDK EV Evals SDK
Organizations
strands-agents strands-labs

strands.vended_plugins.skills.agent_skills

AgentSkills plugin for integrating Agent Skills into Strands agents.

This module provides the AgentSkills class that extends the Plugin base class to add Agent Skills support. The plugin registers a tool for activating skills, and injects skill metadata into the system prompt.

Filesystem skill sources are loaded through the agent’s sandbox (host or container) at init_agent time, not at construction, so each agent sees the skills present on its own filesystem. Skill instances and https:// URLs are sandbox-independent and resolve eagerly at construction.

:meth:Skill.from_url is synchronous, so URLs resolve at construction and no readiness barrier is needed. The observable effect is benign: URL skills are

SkillSource

Section titled “SkillSource”

A single skill source: path string, Path object, or Skill instance.

SkillSources

Section titled “SkillSources”

One or more skill sources.

AgentSkills

Section titled “AgentSkills”
class AgentSkills(Plugin)

Defined in: src/strands/vended_plugins/skills/agent_skills.py:74

Plugin that integrates Agent Skills into a Strands agent.

The AgentSkills plugin extends the Plugin base class and provides:

  1. A skills tool that allows the agent to activate skills on demand
  2. System prompt injection of available skill metadata before each invocation
  3. Session persistence of active skill state via agent.state

Skills can be provided as filesystem paths (to individual skill directories or parent directories containing multiple skills), https:// URLs pointing to raw SKILL.md content, or as pre-built Skill instances.

Filesystem paths are read through the agent’s sandbox at init_agent time, so each agent loads the skills present on its own filesystem (host or container). Skill instances and URLs are sandbox-independent and resolve at construction. As a result, get_available_skills returns filesystem skills only when passed the agent they were loaded for.

Example:

from strands import Agent
from strands.vended_plugins.skills import Skill, AgentSkills


# Load from filesystem
plugin = AgentSkills(skills=["./skills/pdf-processing", "./skills/"])


# Or provide Skill instances directly
skill = Skill(name="my-skill", description="A custom skill", instructions="Do the thing")
plugin = AgentSkills(skills=[skill])


agent = Agent(plugins=[plugin])

__init__

Section titled “__init__”
def __init__(skills: SkillSources,
             state_key: str = _DEFAULT_STATE_KEY,
             max_resource_files: int = _DEFAULT_MAX_RESOURCE_FILES,
             strict: bool = False) -> None

Defined in: src/strands/vended_plugins/skills/agent_skills.py:111

Initialize the AgentSkills plugin.

Arguments:

  • skills - One or more skill sources. Can be a single value or a list. Each element can be:

    • A str or Path to a skill directory (containing SKILL.md)
    • A str or Path to a parent directory (containing skill subdirectories)
    • A Skill dataclass instance
    • An https:// URL pointing directly to raw SKILL.md content

  • state_key - Key used to store plugin state in agent.state.

  • max_resource_files - Maximum number of resource files to list in skill responses.

  • strict - If True, raise on skill validation issues. If False (default), warn and load anyway.

init_agent

Section titled “init_agent”
async def init_agent(agent: Agent) -> None

Defined in: src/strands/vended_plugins/skills/agent_skills.py:144

Initialize the plugin with an agent instance.

Loads any deferred filesystem skill paths through the agent’s sandbox, building the agent’s full skill set. Decorated hooks and tools are auto-registered by the plugin registry.

Arguments:

  • agent - The agent instance to extend with skills support.

skills

Section titled “skills”
@tool(context=True)
async def skills(skill_name: str, tool_context: ToolContext) -> str

Defined in: src/strands/vended_plugins/skills/agent_skills.py:161

Activate a skill to load its full instructions.

Use this tool to load the complete instructions for a skill listed in the available_skills section of your system prompt.

Arguments:

  • skill_name - Name of the skill to activate.
  • tool_context - Injected by the framework. Not user-facing.

get_available_skills

Section titled “get_available_skills”
def get_available_skills(agent: Agent | None = None) -> list[Skill]

Defined in: src/strands/vended_plugins/skills/agent_skills.py:245

Get the list of available skills.

Arguments:

  • agent - When provided, returns that agent’s full skill set (base skills plus filesystem skills loaded from its sandbox). When omitted, returns only the sandbox-independent base skills (Skill instances and URLs); filesystem skills are excluded because they are loaded per-agent at init_agent time.

Returns:

A copy of the resolved skills list.

set_available_skills

Section titled “set_available_skills”
def set_available_skills(skills: SkillSources) -> None

Defined in: src/strands/vended_plugins/skills/agent_skills.py:261

Set the available skills, replacing any existing ones.

Each element can be a Skill instance, a str or Path to a skill directory (containing SKILL.md), a str or Path to a parent directory containing skill subdirectories, or an https:// URL pointing directly to raw SKILL.md content.

Filesystem paths are re-loaded per-agent on the next invocation. Note: this does not persist state or deactivate skills on any agent. Active skill state is managed per-agent and will be reconciled on the next tool call or invocation.

Arguments:

  • skills - One or more skill sources to resolve and set.

get_activated_skills

Section titled “get_activated_skills”
def get_activated_skills(agent: Agent) -> list[str]

Defined in: src/strands/vended_plugins/skills/agent_skills.py:563

Get the list of skills activated by this agent.

Returns skill names in activation order (most recent last).

Arguments:

  • agent - The agent to query.

Returns:

List of activated skill names.