exa-tools
Executes Exa AI search queries via unified Python CLI. Use when searching the web for current information, finding code examples, researching APIs, SDKs, or retrieving programming context for any library or framework.
When & Why to Use This Skill
The Exa Tools skill integrates advanced semantic and neural search capabilities into Claude using the Exa AI API. It allows the agent to move beyond traditional keyword matching to find highly relevant, real-time information, technical documentation, and code context. By leveraging a unified Python CLI, it provides a streamlined way to retrieve up-to-date programming insights, GitHub examples, and API specifications, ensuring the agent's knowledge remains current and accurate.
Use Cases
- Real-time Technical Research: Searching for the latest features, breaking changes, and release notes for modern frameworks like Vite 7 or React 19.
- Code Context & Examples: Finding high-quality code snippets and implementation patterns directly from GitHub to solve specific programming challenges.
- API & SDK Discovery: Researching unfamiliar third-party APIs to understand authentication, endpoints, and integration best practices.
- Semantic Concept Exploration: Using neural search to find deep-dive articles and best practices for complex architectural patterns like Effect-TS or functional programming.
| name | exa-tools |
|---|---|
| type | complex |
| depth | base |
| user-invocable | false |
| description | >- |
[H1][EXA-TOOLS]
Dictum: Semantic search surfaces relevant content.
Execute Exa AI search queries via unified Python CLI.
[IMPORTANT] Commands require --query. API key auto-injected via 1Password.
# Web search (default: auto type, 8 results)
uv run .claude/skills/exa-tools/scripts/exa.py search --query "Vite 7 new features"
# Neural search for concepts
uv run .claude/skills/exa-tools/scripts/exa.py search --query "Effect-TS best practices" --type neural
# Keyword search with more results
uv run .claude/skills/exa-tools/scripts/exa.py search --query "React 19 release notes" --type keyword --num-results 15
# Code context search (GitHub)
uv run .claude/skills/exa-tools/scripts/exa.py code --query "React useState hook examples"
# Code search with custom results
uv run .claude/skills/exa-tools/scripts/exa.py code --query "Effect pipe examples" --num-results 20
[1][OUTPUT]
Commands return: {"status": "success|error", ...}.
| [INDEX] | [CMD] | [RESPONSE] |
|---|---|---|
| [1] | search |
{query: string, results: []} |
| [2] | code |
{query: string, context: []} |