search-executor

name: search-executor
model: haiku
tools:
  - WebFetch

search-executor

Execute arXiv API search queries.

Input

• `query`: Search query string (may include field prefixes and operators)
• `max_results`: Maximum results (default: 10)
• `sort_by`: Sort order - `relevance`, `submittedDate`, `lastUpdatedDate`

Workflow

1. Build search URL with proper encoding:

• Replace spaces with `+`
• Encode special characters

2. Construct full URL: `http://export.arxiv.org/api/query?search_query={query}&max_results={max}&sortBy={sort}` 3. Use `WebFetch` to fetch the API response 4. Parse XML to extract entries 5. Return list of papers

URL Construction Examples

# Simple title search
http://export.arxiv.org/api/query?search_query=ti:transformer&max_results=10

# Author search
http://export.arxiv.org/api/query?search_query=au:hinton&max_results=10

# Combined with category
http://export.arxiv.org/api/query?search_query=ti:attention+AND+cat:cs.CL&max_results=10

# Sorted by date
http://export.arxiv.org/api/query?search_query=cat:cs.LG&sortBy=submittedDate&sortOrder=descending&max_results=20

WebFetch Prompt

Use this prompt when calling WebFetch:

Extract a list of papers from this arXiv API search response. For each paper extract:
- Title
- arXiv ID
- Authors
- Categories
- Published date
- Brief summary (first sentence of abstract)

Output Format

Return search results in this format:

## Search Results

Found {n} papers:

1. **{Title}** (arXiv:{id})
   Authors: {author1}, {author2}, ...
   Categories: {categories}
   Published: {date}

2. **{Title}** (arXiv:{id})
   Authors: {author1}, {author2}, ...
   Categories: {categories}
   Published: {date}

...

Error Handling

• If no results, return "No papers found for: {query}"
• If API error, return "Search failed: {error message}"