Skip to main content

AI glossary Models & Architecture

What is an agent skill?

← All glossary terms
In short

A packaged capability for an AI agent: the instructions, allowed tools, examples and input and output formats it needs to perform one coherent task.

Explained

An agent skill bundles everything a model needs to perform a defined capability reliably and repeatedly, for example filing an expense or running a compliance check. It typically contains instructions written for the agent, the tools it may use, examples, scripts or reference files, and the expected inputs and outputs. Anthropic formalised the idea as Agent Skills: folders with a SKILL.md file that an agent discovers by name and description and loads only when a task calls for it. Skills are the unit of reuse between specifications, runbooks and orchestration layers such as Synapse.

What goes inside a skill

  • A name and a short description, so the agent knows when the skill applies.
  • Instructions: the procedure, rules and quality criteria for the task.
  • Tools and scripts: the actions the skill may take, with their parameters.
  • Reference material: templates, examples, schemas or domain documents.
  • Checks: how to verify the output before returning it.

How agents use skills

An agent does not read every skill up front. At start it sees only each skill's name and description. When a request matches, it loads the skill's instructions, and only then opens the extra files it needs. Anthropic calls this progressive disclosure: an agent can have many skills available without filling its context window.

This makes skills a practical way to give a general agent specialised know-how, much like an onboarding guide for a new team member.

Skills, tools and prompts

A tool is a single action the agent can call, such as "search the CRM" or "send an email". A prompt is an instruction for one interaction. A skill sits above both: it packages the procedure, the tools and the knowledge needed for a whole task, so the same capability behaves consistently every time and can be versioned, reviewed and shared.

In an organisation, skills become a catalogue of reusable capabilities that different agents and teams can draw on.

Frequently asked questions

What are Agent Skills in Claude?

Agent Skills are Anthropic's format for packaging capabilities for Claude: a folder with a SKILL.md file containing a name, a description and instructions, plus optional scripts and resources. Claude loads a skill when a task matches its description.

What is the difference between an agent skill and a tool?

A tool is one action, like querying a database. A skill is a packaged procedure that can use several tools, with the instructions and knowledge to do a complete task well. Tools are the verbs; a skill is the know-how of when and how to use them.

What is a SKILL.md file?

It is the main file of an Agent Skill. It starts with metadata, at least a name and a description, followed by the instructions the agent follows when the skill is triggered. It can point to other files in the same folder that the agent reads only if needed.

Why use skills instead of one long system prompt?

A long prompt loads everything into every conversation, costs context and is hard to maintain. Skills load only when relevant, can be versioned and tested one by one and can be reused across agents.

How do agent skills relate to MCP?

They are complementary. MCP (Model Context Protocol) connects an agent to external tools and data sources. A skill tells the agent how to carry out a task, which may include using tools exposed through MCP.

Where it shows up at Thinkia

Products

Further reading

Sources

Updated: 2026-10-21

Want to see how this applies to your organisation?

Talk to an AI Expert