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

# Skills

> Install Surnex as a reusable skill for AI agents.

You can install Surnex as a skill for AI agents that support the `skills` ecosystem. It gives your agent a versioned workflow guide for operating Surnex instead of leaving it to work out how the pieces fit together each session.

## Install

```bash theme={null}
npx skills add surnex/skills --skill surnex
```

Restart your AI tool afterwards if it doesn't pick the skill up immediately.

## What the skill covers

The `surnex` skill helps your agent:

* Understand that **most tools read stored data**, not live results — reading rankings doesn't check rankings, and a figure can legitimately be a day old
* Know that **creating a project collects nothing immediately**: it creates four schedules and the first data can be 24 hours away
* Know that **audits, web vitals, domain analysis, and keyword research never run on a schedule** and have to be started explicitly
* Recognise **which tools spend plan allowance**, and avoid the failure mode that matters — looping a billable tool over a keyword list
* Handle **organization ambiguity**: with several memberships the server refuses to guess rather than billing the wrong one
* Understand it **acts as you**, carrying your role, so a member's connection can't create or delete projects
* Pass **location and language explicitly**, since tools default to the United States rather than inheriting your project's market

It also names the behaviours that read as failures but aren't: an empty new project waiting on its first scheduled run, a not-found that means wrong organization rather than deleted, and a second audit refused because one is already crawling.

Alongside the guide, it ships references for [all 88 tools](/mcp/tools) flagged for writes and cost, worked [workflows](/mcp/playbooks), how to read the numbers without misinterpreting them, and troubleshooting.

## Why use a skill

Without one, your agent works Surnex out from tool descriptions alone. A description says what one tool does; it can't say what order to do things in, what a call costs, or which empty result is a finding rather than a fault.

With a skill you get:

* A stable reference for how the product fits together
* Use-case guidance instead of tool-by-tool guessing
* Explicit caveats — what's asynchronous, what's irreversible, what's billable
* A versioned source of truth that updates as Surnex changes

## Repository

The public source lives at [github.com/surnex/skills](https://github.com/surnex/skills).

## Notes

* Pair it with the [MCP server](/mcp/connect) — the skill explains the workflows, the connector executes them
* The skills ecosystem is supported by Claude Code and claude.ai. Through another client, the MCP server still ships the essentials in its own tool descriptions
* The skill assumes [OAuth](/mcp/auth). It won't help an agent configured with an API key, which the MCP endpoint refuses