Exa
Shanone’s Exa integration gives your agent 12 tools for neural/semantic search — going beyond keyword search with similarity search, direct answer generation, deep multi-step research, and asynchronous research agents.Getting Started
1
Get an Exa API key
Create an API key from the Exa dashboard.
2
Add the key in Shanone
Open Integrations in the Shanone dashboard, select Exa, and paste the API key.
3
Retry your request
Once saved,
shanone_execute_tool calls for exa_* tools will succeed.Available Tools
Shanone provides 12 tools for Exa, organized into these categories:Search & content
Search & content
exa_search, exa_find_similar, exa_get_contents, exa_answer, exa_deep_searchResearch agents
Research agents
exa_agent_create, exa_agent_get, exa_agent_list, exa_agent_cancel, exa_agent_delete, exa_agent_eventsCommon Use Cases
Grounded question answering
Use
exa_answer to get a direct answer with cited sources instead of a raw results listCompetitive research
Find pages similar to a competitor’s site with
exa_find_similar, then pull full content with exa_get_contentsMulti-step deep research
Kick off
exa_deep_search or a research agent for a complex query that needs iterative explorationLong-running background research
Create a research agent with
exa_agent_create and poll exa_agent_events while it runs asynchronouslyTroubleshooting
exa_search results feel too broad or off-topic
exa_search results feel too broad or off-topic
Exa ranks by semantic meaning rather than keywords — phrase the query as a natural-language description of what you want, not a keyword string, and narrow with domain/date filters where supported.
A research agent seems stuck in progress
A research agent seems stuck in progress
Long-running research jobs created with
exa_agent_create are asynchronous; poll exa_agent_get or exa_agent_events for status rather than expecting a synchronous result, and use exa_agent_cancel if it needs to be stopped.exa_get_contents returns empty or truncated content
exa_get_contents returns empty or truncated content
Some pages block scraping or require JavaScript rendering that Exa’s content fetcher can’t execute; fall back to
exa_search snippets or try exa_find_similar for an alternate source.