Exa

View saved version →

---
name: Search
description: "Deep research powered by Exa. Use for lead generation, literature reviews, deep dives, competitive analysis, or any query where one search falls short, including phrases like 'research this', 'find everything about', 'find me all', or 'deep dive on'."
---

# Exa Search And Research

Use Exa to search the web and fetch source pages when the user needs current web information, research support, source discovery, or source-grounded synthesis. Understand the query, plan the searches, validate results, and deliver a compact answer with useful links.

This skill uses the bundled Exa app tools:
- `web_search_exa` for web search
- `web_fetch_exa` for reading known URLs

## Prerequisites: Exa App Connection

The bundled Exa app must be connected before this skill can use Exa tools.

If `web_search_exa` or `web_fetch_exa` is unavailable:
1. Ask the user to connect or re-authenticate the Exa app from the plugin/app install flow.
2. Tell them to start a new Codex thread if the tools still do not appear after connecting.
3. Do not fall back to generic web search, ask the user to re-try connecting to the Exa App.

On auth, connection, or rate-limit errors, surface the Exa app connection fix clearly.

## Date Calculation (Do This First)

If the query involves time ("last week", "recent", "past 6 months"), calculate exact dates from today's date in your environment context. Write out the calculation explicitly before doing anything else. Never eyeball dates or reuse dates from examples.

## Step 1: Assess the Query

Read the user's query and determine two things:

**How complex is this?**
- **Extremely Simple** (e.g. reading the contents of 1-2 pages): Handle it yourself. Read `references/searching.md` for query-writing guidance, run the searches, review and filter results, then respond directly.
- **Moderate** (when a fast or low-effort search is requested): Run a small set of targeted searches and fetch only the most useful pages.
- **Advanced** (clear topic, clear filters, a few parallel searches): Break the task into distinct search angles, search each angle, then compile.
- **Complex** (cross-referencing across entity types, multi-hop chains, exhaustive coverage, semantic filtering): Use a multi-pass workflow where later searches depend on earlier findings.

**Confirm when ambiguous:**
If the query could reasonably be handled as Extremely Simple/Moderate OR as Advanced/Complex, pause and ask the user before proceeding. Present:
1. Your interpretation of the query
2. The two (or more) plausible complexity levels
3. What each level would look like in practice (e.g., "I can do a quick 1-2 search lookup, or I can do a deeper multi-angle sweep")
4. Let the user choose

Examples of ambiguous queries:
- "What are the best LLM fine-tuning frameworks?" — could be a quick opinionated list (Moderate) or an exhaustive evaluated comparison (Complex)
- "Find competitors to Acme Corp" — could be a quick search for known competitors (Moderate) or a deep sweep across funding databases, press, and niche directories (Complex)
- "What's the latest on WebGPU?" — could be one news search (Extremely Simple) or a multi-angle survey of specs, browser support, community adoption, and benchmarks (Advanced)

Do NOT ask for confirmation when:
- The query is clearly extremely simple (fact lookups, single-entity questions)
- The query is clearly complex (explicit multi-constraint, "find everything", "exhaustive", "comprehensive")
- The user has already specified depth ("do a deep dive", "quick answer")

Note: if the user explicitly asks for something (e.g. "100" of something), continue to work until you've achieved it.

**What work needs to happen?** Identify which of these apply (most queries use 3-5):

1. **Seed from user input**: The user provided a list of entities to start from (company names, tickers, paper titles). Each seed becomes a parallel workstream.
2. **Define what qualifies**: What makes a result a valid "row"? Translate t

Preview of the first 4,000 characters. Read full instructions →

Referenced files: 11 — file references are metadata, not package contents.