exa-search

name	exa-search
description	Semantic search using Exa API for security research queries. Unlike keyword-based search, Exa understands context and retrieves high-quality, relevant results for technical research.
origin	openclaw
version	0.2.0.2
compatibility	["openclaw","claude-code","cursor","windsurf"]
allowed-tools	["Bash","Read","Write","Edit","WebSearch","WebFetch","Agent"]
defense_triple_required	false
metadata	{"domain":"research","tool_count":0,"guide_count":5,"last_reviewed":"2026-07-26"}

Exa Search

Summary

Unlike keyword-based search, Exa understands context and retrieves high-quality, relevant results for technical research.

Domain: research

Skill Identity

Attribute	Value
Domain	Intelligence Gathering
Skill ID	exa-search
Version	1.0.0
Hacker Laws	Law 3 (Intelligence Over Force), Law 9 (Systematic Over Random)
Related Skills	deep-research, osint, social-intelligence

Purpose

Semantic search using Exa API for security research queries. Unlike keyword-based search, Exa understands context and retrieves high-quality, relevant results for technical research.

Core Capabilities

1. Semantic Search: Context-aware query understanding
2. Date Filtering: Recent content prioritization
3. Domain Filtering: Target specific sources
4. Content Extraction: Full-text retrieval

Use Cases

• CVE Research: "Recent CVEs affecting Spring Boot applications"
• Exploit Techniques: "SSRF bypass techniques in AWS metadata service"
• Tool Research: "Best tools for JWT security testing"
• Threat Intelligence: "APT campaigns targeting healthcare 2024"

Exa API Reference

• Endpoint: https://api.exa.ai/search
• Auth: API key via X-API-Key header
• Docs: https://docs.exa.ai/

Query Strategy

• Semantic phrasing: Frame queries as full sentences ("How does X bypass Y?") rather than keyword strings; Exa's embedding model rewards natural-language intent.
• Iterative narrowing: Start broad, then re-query with terms surfaced by the first batch to drill into specifics.
• Source class filters: Restrict to includeDomains (e.g., ["github.com","arxiv.org"]) for code/research, or excludeDomains for noisy aggregators.