# Agent Skills

> Use the Search1API Agent Skill bundled with the Search1API Claude plugin, or install the standalone CLI skill for a terminal agent.

Canonical URL: https://s1.dev/docs/integrations/skills

An MCP server exposes tools an agent can call. A skill provides instructions for choosing those tools, reading sources, and answering with citations. Search1API provides a skill with the published Claude plugin and a separate standalone skill for agents that use the `s1` CLI.

## Install

### Claude plugin

For Claude apps, Cowork, and Claude Code, [install Search1API plugin](https://claude.ai/customize/plugins/id/c73feffb-aae4-49f4-8515-24322b313989%40anthropic-plugin-directory). You can also find it under **Customize → Plugins → Discover** by searching for **Search1API**. The published plugin bundles an Agent Skill with the remote MCP tools. Connect Search1API through OAuth after adding the plugin. Plugins require a paid Claude plan.

For the same tools without the bundled skill, [add Search1API connector](https://claude.ai/directory/search1api). See the [Claude guide](https://s1.dev/docs/integrations/claude) for installation steps and Claude Code account sync, or [MCP setup](https://s1.dev/docs/integrations/mcp) for manual configuration.

### Standalone CLI skill

For terminal agents that use the `s1` CLI:

```bash
npm install -g search1api-cli
npx skills add superagents-lab/search1api-cli
s1 login
```

This standalone skill drives the `s1` CLI, so the CLI has to be installed and logged in first. If it isn't, the skill will walk the user through it. The Claude plugin uses MCP tools directly instead.

## What the skill guides

Both skills provide guidance for adapting web access to the request:

- Read a shared URL with `crawl`.
- Use a small result set for a quick lookup; for a research question, search more broadly and read relevant pages for full text.
- Apply a recency filter for requests such as "latest" or "today".
- Combine tools for research, URL summarization, and trending-topic exploration.
- Synthesize findings and cite source URLs.

The Claude plugin skill uses MCP parameters such as `max_results` and `time_range`. The standalone skill uses CLI flags such as `-n` and `-t`.

## Using it

Once installed, just talk to your agent normally:

- "search for the latest AI news"
- "what does this link say? https://example.com"
- "what's trending on GitHub?"
- "research quantum computing thoroughly"

Claude or your agent uses the skill's instructions to choose a tool or command, the search service, the result count, and whether to read pages.

## Cost

Both skills use the same Search1API account credits as direct API calls. The skill itself adds no separate Search1API fee; workflows that make more searches or read more pages use more credits. See [Credits and limits](https://s1.dev/docs/essentials/credits-and-limits).

Sources: [Claude plugin skill](https://github.com/superagents-lab/search1api-mcp/tree/main/claude-plugin/skills/search1api) · [Standalone CLI skill](https://github.com/superagents-lab/search1api-cli/tree/master/skills/search1api)