home / skills / vaayne / agent-kit / mcp-exa-search

mcp-exa-search skill

needs review

This skill helps you perform web searches, company research, and code context lookups using Exa AI for fresh, relevant results.

npx playbooks add skill vaayne/agent-kit --skill mcp-exa-search

Review the files below or copy the command above to add this skill to your agents.

Files (1)

SKILL.md

3.0 KB

---
name: mcp-exa-search
description: Search the web and find code examples using Exa AI. Use when performing web searches, finding current information, or looking up code examples and documentation. Triggers on "search the web", "find online", "web search", "search for", "find code examples", "Exa search".
---

# Exa AI Search

MCP service at `https://mcp.exa.ai/mcp` (http) with 2 tools.

## Requirements

- `mh` CLI must be installed. If not available, install with:
  ```bash
  curl -fsSL https://raw.githubusercontent.com/vaayne/mcphub/main/scripts/install.sh | sh
  ```

## Usage

```
List tools: `mh list -u https://mcp.exa.ai/mcp -t http`
Get tool details: `mh inspect -u https://mcp.exa.ai/mcp -t http <tool-name>`
Invoke tool: `mh invoke -u https://mcp.exa.ai/mcp -t http <tool-name> '{"param": "value"}'`
```

## Notes

- Run `inspect` before invoking unfamiliar tools to get full parameter schema
- Timeout: 30s default, use `--timeout <seconds>` to adjust
- `getCodeContextExa` is best for programming-related queries (libraries, SDKs, APIs)
- `webSearchExa` returns clean text content ready for LLM use

## Tools

| Tool                | Description                                                                                                                                                                        |
| ------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `webSearchExa`      | Search the web for any topic and get clean, ready-to-use content. Best for: current information, news, facts, or answering questions about any topic.                              |
| `getCodeContextExa` | Find code examples, documentation, and programming solutions. Searches GitHub, Stack Overflow, and official docs. Best for: API usage, library examples, code snippets, debugging. |

## Tool Parameters

### `webSearchExa`

```
Required:
  query (string)                — web search query

Optional:
  numResults (number)           — number of results to return (default: 8)
  category (string)             — filter: "company" | "research paper" | "people"
  livecrawl (string)            — "fallback" (default) | "preferred"
  type (string)                 — "auto" (default) | "fast"
  contextMaxCharacters (number) — max chars in response (default: 10000)
```

### `getCodeContextExa`

```
Required:
  query (string)         — search query, e.g. "React useState hook examples"

Optional:
  tokensNum (number)     — tokens to return, 1000–50000 (default: 5000)
```

## Examples

```bash
# Web search
mh invoke -u https://mcp.exa.ai/mcp -t http webSearchExa '{"query": "openclaw plugins"}'

# Web search with options
mh invoke -u https://mcp.exa.ai/mcp -t http webSearchExa '{"query": "Go 1.23 release notes", "numResults": 5}'

# Code context
mh invoke -u https://mcp.exa.ai/mcp -t http getCodeContextExa '{"query": "Go chi router middleware examples"}'
```

Overview

This skill provides web search, company research, and code-context lookup using Exa AI. It exposes three focused tools for real-time web results, deep company profiles, and programming documentation context. Use it to augment research, gather fresh information, and pull relevant code examples or API details for development tasks.

How this skill works

The skill connects to an MCP HTTP service that offers three tools: webSearchExa, companyResearchExa, and getCodeContextExa. Each tool accepts a JSON request and returns structured results drawn from current web content, news, financials, and programming documentation. Inspect tool schemas before invoking to understand parameters, and adjust timeouts for longer queries.

When to use it

Gather up-to-date web results on a topic or news event.
Compile a detailed company profile, including operations, news, and financial signals.
Fetch code examples, API docs, or library-specific context for development tasks.
When you need configurable result counts or focused sources for research.
When you require fresh, real-time context that a static dataset may not contain.

Best practices

Run the tool inspection command first to see available parameters and expected payloads.
Use getCodeContextExa for programming queries to get higher-quality, code-focused context.
Limit result counts for faster responses; increase timeout for deeper research.
Provide precise queries (library name, API method, company name) to improve relevance.
Validate returned snippets and follow links to original sources for verification.

Example use cases

Search the web for the latest AI developments or policy changes affecting your product.
Research a potential partner or competitor to gather news, financials, and market position.
Find API usage examples and implementation notes for a specific library or SDK.
Collect citations and source links for a report or due-diligence summary.
Quickly retrieve code patterns or migration tips when upgrading dependencies.

FAQ

What tools are available and when should I pick each?

Use webSearchExa for general web queries, companyResearchExa for business-focused investigations, and getCodeContextExa for programming documentation and code examples.

How do I run queries against the service?

Invoke tools by sending JSON payloads to the MCP HTTP endpoint. Inspect a tool first to see its parameter schema, then call it with the appropriate payload and a suitable timeout.