Pi Exa
Extension
Web search and URL content extraction via Exa, giving the agent source-backed discovery and bounded page reads.
By Vedang Manerikar · git:github.com/vedang/pi-exa · View source
What it does
Pi Exa gives the harness two focused web research tools: exa_search for finding current sources on the open web, and exa_contents for pulling highlights or bounded text from specific URLs you already know.
Responses return structured titles, URLs, publication dates, and excerpts rather than raw HTML dumps. Oversized outputs are truncated in context and written to private temporary files (0600) so the agent can inspect them selectively without blowing the token budget.
When to reach for it
Invocation: The agent calls exa_search or exa_contents when a task needs external docs, release notes, or live web pages.
- Checking current library docs or upstream issues
- Run exa_search with domain or date filters, then pull exact sections with exa_contents.
- Reading a URL pasted in chat
- Fetch clean text or highlights directly with exa_contents without running a broad search first.
What it adds
- exa_search tool with category, domain, date, and search-type filters.
- exa_contents tool for extracting highlights or full text from up to 10 URLs at once.
- Automatic EXA_API_KEY redaction and private temp-file spillovers for large responses.
How it fits
- Chrome CDP — Use Pi Exa first for fast, headless public web lookups; reach for Chrome CDP only when a page requires a logged-in browser session or live DOM inspection.
Good to know
- Requires EXA_API_KEY in the shell environment; if missing, the tools return a clear configuration error instead of crashing the session.
Source and setup
- Loaded via
git:github.com/vedang/pi-exa- Attribution
- By Vedang Manerikar