> ## Documentation Index
> Fetch the complete documentation index at: https://handler.alduncanson.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent skill

> Install the Handler skill so Claude Code, Cursor, and other coding agents know how to drive Handler without guessing.

# Give agents a Handler skill

These docs are published as an agent skill: a `SKILL.md` file in the open
[Agent Skills](https://agentskills.io) format that coding agents load on demand.
It is generated from the documentation on every deploy, so it always describes
the current CLI, including the auth schemes, attachment flags, and task commands.

```text theme={null}
https://handler.alduncanson.com/skill.md
```

The site also serves a discovery index at
`https://handler.alduncanson.com/.well-known/skills/index.json`, which is how
skill installers find it from the base URL.

## Install with the skills CLI

The [skills CLI](https://github.com/vercel-labs/skills) discovers the skill
from the docs URL, detects the agents installed on your machine (Claude Code,
Cursor, Windsurf, and others), and lets you install to all of them or pick:

```bash theme={null}
npx skills add https://handler.alduncanson.com
```

Run it again whenever you want the latest version; the skill changes when the
docs change.

## Install by hand

Claude Code loads skills from `~/.claude/skills/<name>/SKILL.md` for every
project, or from `.claude/skills/<name>/SKILL.md` inside one repository:

```bash theme={null}
mkdir -p ~/.claude/skills/handler
curl -fsSL https://handler.alduncanson.com/skill.md -o ~/.claude/skills/handler/SKILL.md
```

Other agents that read the same format take the file in their own skills
directory.

## What the skill teaches

* when Handler is the right tool, and when it is not
* the core commands with their flags: cards, messages, attachments, tasks,
  sessions, push notifications, server configuration
* the six authentication schemes, how each is configured, and that credentials
  come from environment variables and only go to the origin you configured
* structured output (`--output json` and `--output ndjson`) for scripting
* a step-by-step workflow from adding a server to bridging Handler into MCP
* the gotchas that trip people up, and a verification checklist

## Pair it with live docs

A skill is a snapshot of what Handler can do. For questions the skill does not
answer, give the agent the hosted [documentation MCP](/guides/agent-docs-mcp)
at `https://handler.alduncanson.com/mcp`, or point it at
`https://handler.alduncanson.com/llms.txt` for the full page index.

## Recommended agent instruction

```text theme={null}
Use the Handler skill to talk to A2A agents. Prefer named servers from
servers.toml over raw URLs, keep credentials in environment variables, and use
`handler --output json` when another program or agent will consume the result.
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.