> For the complete documentation index, see [llms.txt](https://docs.eyepop.ai/developer-documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.eyepop.ai/developer-documentation/agent-skills.md).

# Agent Skills

Install the eyepop skill for your preferred agent harness

The `eyepop` skill teaches your coding agent to drive EyePop — install the CLI, sign in, run inference, create abilities, evaluate datasets, and write Python or Node SDK code.

It ships in the [Agent Skills](https://agentskills.io) format, so it is not tied to one agent. Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, Windsurf, Cline, and any other harness that reads the format can load it.

### Install

The default install detects the agents in your project and asks where to put the skill:

```bash
npx skills add eyepop-ai/eyepop-skills
```

Install for a single agent:

```bash
npx skills add eyepop-ai/eyepop-skills -a claude-code
```

Install for every agent it detects:

```bash
npx skills add eyepop-ai/eyepop-skills -a '*'
```

Install globally, so the skill is available in every project instead of just this one:

```bash
npx skills add eyepop-ai/eyepop-skills -g
```

Preview what the repo ships without installing anything:

```bash
npx skills add eyepop-ai/eyepop-skills --list
```

Without `npx`, copy `skills/eyepop` from the repository into your agent's skills directory — for example `.claude/skills/eyepop` for Claude Code.

### Verify it loaded

Start a session in your project and ask:

> What EyePop skills do you have loaded?

The agent names the `eyepop` skill and summarizes what it covers.

### Your first inference

Once the skill is installed, hand your agent this prompt and it takes over from there:

> Use your eyepop skill to install the eyepop cli on this system and walk the user through their first inference

You need an EyePop account with an active plan and an API key (`eyp_...`) from the [dashboard](https://dashboard.eyepop.ai) — see [API Keys](/developer-documentation/api-keys.md) and [Pricing](/developer-documentation/pricing.md).

### What the agent can do with it

* Check the machine, install the CLI for your OS, and sign in
* List your models, abilities, Pops, deployments, instances, and datasets, then prove a run works
* Run inference on files, directories, and URLs, and parse the JSON result
* Write a prompt, create an ability, test it on real media, iterate, and give it an alias
* Build a dataset, add ground truth, evaluate an ability, and read the metrics
* Create, patch, and delete persistent deployments
* Set up and operate an on-premise instance, and route runs to it
* Write Python or Node SDK code: sessions, sources and options, composable Pops, results, the data endpoint

[View on GitHub](https://github.com/eyepop-ai/eyepop-skills)