Skills

A skill is a named Markdown fragment (an instruction or workflow) that an agent loads during a session when its description matches the task. Use a built-in skill or create your own, then attach it to an agent by name.

Field	Required	Description
`name`	Yes	How agents reference the skill.
`description`	Yes	When to use the skill. The agent sees only this before deciding to load the `body`.
`body`	Yes	The Markdown instructions the agent loads.
`source`	No	Where the content came from.
`url_pattern`	No	A hint at URLs where the skill applies. Shown to the agent, never matched against the live page.

Exact constraints: Create a skill.

Create and attach a skill

curl -X POST https://agp.eu.hcompany.ai/api/v2/skills \
  -H "Authorization: Bearer $HAI_API_KEY" -H "Content-Type: application/json" \
  -d '{
    "name": "extract-table-data",
    "description": "Use when the task asks for data shown in an HTML table.",
    "body": "Extract all rows and columns into a JSON array of objects, using the headers as keys."
  }'

curl -X POST https://agp.eu.hcompany.ai/api/v2/agents \
  -H "Authorization: Bearer $HAI_API_KEY" -H "Content-Type: application/json" \
  -d '{
    "name": "table-scraper",
    "description": "Scrapes structured data from web tables.",
    "environments": ["h/browser"],
    "skills": ["extract-table-data"]
  }'

from hai_agents import Client

client = Client()
client.skills.create_skill(
    name="extract-table-data",
    description="Use when the task asks for data shown in an HTML table.",
    body="Extract all rows and columns into a JSON array of objects, using the headers as keys.",
)
client.agents.create_agent(
    name="table-scraper",
    description="Scrapes structured data from web tables.",
    environments=["h/browser"],
    skills=["extract-table-data"],
)

import { HaiAgentsClient } from "hai-agents";

const client = new HaiAgentsClient();
await client.skills.createSkill({
  name: "extract-table-data",
  description: "Use when the task asks for data shown in an HTML table.",
  body: "Extract all rows and columns into a JSON array of objects, using the headers as keys.",
});
await client.agents.createAgent({
  name: "table-scraper",
  description: "Scrapes structured data from web tables.",
  environments: ["h/browser"],
  skills: ["extract-table-data"],
});

Write effective skills

Make the description a trigger. It is all the agent sees before loading, so phrase it as “use when…” and name the situation precisely. Vague descriptions load at the wrong time or never.
Keep the body self-contained. The agent reads it cold, mid-task: write a focused procedure (steps, formats, edge cases), not background.
Put always-on behavior in the agent’s instructions. Skills are for procedures that apply only some of the time.
Compose small skills. Narrow skills route better than one broad one.

Built-in skills

H maintains computer-use skills under the h/ namespace. Attach them to any agent by name. List them with GET /api/v2/skills?name=h/ (List skills), or fetch one with GET /api/v2/skills/h/{skill_name} (Retrieve a skill).

How an agent uses a skill

Attached skills are listed in the system prompt by name and description only (an <available_skills> section, with url_pattern as a hint when set). When a description looks relevant, the agent calls the built-in load_skill tool with that name, and the body enters the conversation as instructions for the next steps. A wrong name returns the closest available names, so the agent can retry. Environments bundle their own skills. Those are always in context, fixed per environment, and not configurable.

Get started

Concepts

Browser and desktop

Integrations

Guides

Operate

Create and attach a skill

Write effective skills

Built-in skills

How an agent uses a skill

​Create and attach a skill

​Write effective skills

​Built-in skills

​How an agent uses a skill

Create and attach a skill

Write effective skills

Built-in skills

How an agent uses a skill