web-search

name	web-search
description	Search web content using Exa advanced search across categories - personal sites/blogs, research papers, and tweets/X. Use when searching for web content, academic papers, blog posts, or social media discussions.

Web Search (Exa Advanced)

Tool Restriction

ONLY use web_search_advanced_exa. Select the category parameter based on search type: "personal site", "research paper", or "tweet".

Token Isolation

Never run Exa searches in main context. Always spawn Task agents:

• Agent calls web_search_advanced_exa with the appropriate category
• Agent merges and deduplicates results before presenting
• Agent returns distilled output (brief markdown or compact JSON)
• Main context stays clean regardless of search volume

Personal Sites & Blogs

Category: "personal site". Full filter support.

Use when you need:

• Individual expert opinions and experiences
• Personal blog posts on technical topics
• Portfolio websites
• Independent analysis (not corporate content)
• Deep dives and tutorials from practitioners

Example - technical blog posts:

web_search_advanced_exa {
  "query": "building production LLM applications lessons learned",
  "category": "personal site",
  "numResults": 15,
  "type": "deep",
  "enableSummary": true
}

Example - exclude aggregators:

web_search_advanced_exa {
  "query": "startup founder lessons",
  "category": "personal site",
  "excludeDomains": ["medium.com", "substack.com"],
  "numResults": 15,
  "type": "auto"
}

Research Papers

Category: "research paper". Full filter support.

Use when you need:

• Academic papers from arXiv, OpenReview, PubMed, and similar sources
• Scientific research on specific topics
• Literature reviews with date filtering
• Papers containing specific methodologies or terms

Example - recent papers:

web_search_advanced_exa {
  "query": "transformer attention mechanisms efficiency",
  "category": "research paper",
  "startPublishedDate": "2024-01-01",
  "numResults": 15,
  "type": "auto"
}

Example - specific venues:

web_search_advanced_exa {
  "query": "large language model agents",
  "category": "research paper",
  "includeDomains": ["arxiv.org", "openreview.net"],
  "includeText": ["LLM"],
  "numResults": 20,
  "type": "deep"
}

Tweets & X Content

Category: "tweet". LIMITED filter support.

Critical: The following parameters are NOT supported for tweets and will cause errors (400 or 500):

• includeText - NOT SUPPORTED
• excludeText - NOT SUPPORTED
• includeDomains - NOT SUPPORTED
• excludeDomains - NOT SUPPORTED
• moderation - NOT SUPPORTED (causes 500 server error)

Put keyword constraints in the query string instead of text/domain filters. Use date filters to narrow results.

Use when you need:

• Social discussions on a topic
• Product announcements from company accounts
• Developer opinions and experiences
• Trending topics and community sentiment
• Expert takes and threads

Example - recent tweets:

web_search_advanced_exa {
  "query": "Claude Code MCP experience",
  "category": "tweet",
  "startPublishedDate": "2025-01-01",
  "numResults": 20,
  "type": "auto",
  "livecrawl": "preferred"
}

Example - keywords in query (not includeText):

web_search_advanced_exa {
  "query": "launching announcing new open source release",
  "category": "tweet",
  "startPublishedDate": "2025-12-01",
  "numResults": 15,
  "type": "auto"
}

Common Parameters

Parameters shared across all categories:

Core

• query (required)
• numResults
• type - "auto", "fast", "deep", "neural"

Date filtering (ISO 8601)

• startPublishedDate / endPublishedDate
• startCrawlDate / endCrawlDate

Content extraction

• textMaxCharacters / contextMaxCharacters
• enableSummary / summaryQuery
• enableHighlights / highlightsNumSentences / highlightsPerUrl / highlightsQuery

Additional (personal site and research paper only)

• includeDomains / excludeDomains
• includeText / excludeText - single-item arrays only; multi-item arrays (2+ items) cause 400 errors; use query for multiple terms
• additionalQueries
• livecrawl / livecrawlTimeout
• subpages / subpageTarget

Output Format

For all categories, return:

1. Results - title, author/source, date, key insights or summary
2. Sources - URLs with context (venue for papers, handle for tweets)
3. Notes - relevant observations (author expertise, methodology, sentiment)

Related

• skills/code-search/ — Use for code snippets, API examples, and SDK usage via get_code_context_exa