AgentTools.<tool>(tool_name="...", ...) or agent.add_tool.<tool>(tool_name="...", ...). Every tool requires a unique tool_name=. Each entry lists the tool’s configuration parameters. See the Agent reference for attaching and running tools.
api
Make an API request to a given URL.
Platform docs: API
Whether to return the raw JSON response from the API
Target URL for the API Request
The body parameters to include in the API request
Files to include in the API request
Headers to include in the API request
Choose the API Method desired (GET, POST, PUT, DELETE, PATCH)
Query parameters to include in the API request
The raw JSON request to the API
arxiv
Query ARXIV to return relevant articles
Platform docs: Arxiv Loader
Whether to chunk the text
The ARXIV query
The overlap of the chunks
The size of the chunks to create
crunchbase
Call the Crunchbase API to look up companies, people, funding rounds, and acquisitions.
Platform docs: Crunchbase
JSON body for searches/* endpoints. IMPORTANT: categories and location_identifiers require UUIDs from autocomplete, not text! Example: {“field_ids”:[“identifier”,“funding_total”],“query”:[{“type”:“predicate”,“field_id”:“categories”,“operator_id”:“includes”,“values”:[“uuid-from-autocomplete”]}],“limit”:25}
API path: ‘autocompletes’ | ‘entities/organizations/{id}’ | ‘entities/people/{id}’ | ‘searches/organizations’ | ‘searches/funding_rounds’ | ‘searches/people’
Query string for GET requests. Autocomplete: ‘query={name}&collection_ids=organization.companies’. Entity lookup: ‘field_ids=short_description,funding_total&card_ids=founders’
exa_ai
Query the Exa search API
Platform docs: Exa AI Search
Only return results crawled before this date (ISO 8601)
Only return results published before this date (ISO 8601, e.g. 2024-12-31T00:00:00.000Z)
Enable live crawling to fetch fresh content. When on, always fetches live content. When off, uses cached content.
Select the search category to match the user’s query intent. Default to EXA_AI_SEARCH: General web search. Use general search for weather
Maximum characters per result for text or highlights content
Number of results to return (1-100) default 10
The search query
Search method: auto (default) intelligently combines methods, neural uses embeddings, fast is streamlined, instant is lowest latency, deep/deep-reasoning/deep-max offer increasing search depth, Generally default to auto
Only return results crawled after this date (ISO 8601)
Only return results published after this date (ISO 8601, e.g. 2024-01-01T00:00:00.000Z)
Return LLM-selected highlights instead of full text. Uses ~10x fewer tokens with better relevance.
google_search
Query the Google Search search API
Platform docs: Google Search
The location of the search
The number of results to return
The Google search query
Select the search type: Web, Image, Hotels, Events, or News
parallel_ai_search
Search the web with Parallel’s Search API
Platform docs: Parallel Search
Exclude these domains from the response.
Limit the excerpt length returned inside the excerpts object.
Enable to fetch the latest version of each page instead of relying on cached index entries.
Legacy field for truncating excerpts. Prefer the excerpts.* controls when available.
Upper bound on the number of results to return.
Choose output presets. one-shot maximizes coverage, agentic optimizes for iterative agent loops.
Describe what the search should accomplish. Provide preferred sources or freshness hints here.
Prioritize these domains in ranked results.
Legacy tier selector. Leave blank to use the default tier for your workspace.
Optional structured keyword queries. At least one of objective or search queries is required.
perplexity_search
Query the Perplexity search API
Platform docs: Perplexity Search
Filter results last updated after this date
Filter results last updated before this date
The maximum number of results to return
Maximum tokens per page
The search query
Additional queries to search for
Whether to return images in results
Whether to return snippets in results
Filter results published after this date
Filter results published before this date
Filter results to specific domains
The search mode to use
Filter results by recency
The latitude of the user’s location
The longitude of the user’s location
The name of the user’s location
The radius in kilometers for location-based search
url_loader
Scrape the contents from a URL
Platform docs: URL Scraper: Scrape URL
The provider to use for the URL Scraper
One of:
apify, jina, modalThe API key to use
The URL to load
Whether to enhance the content
Use a proxy to crawl the website
wikipedia
Query Wikipedia to return relevant articles
Platform docs: Get Wikipedia Article
Whether to chunk the text
The Wikipedia query
The overlap of the chunks
The size of the chunks to create
youtube
Get the transcript of a youtube video.
Platform docs: Get YouTube Transcript
Whether to chunk the text
The YouTube URL to get the transcript of
The overlap of the chunks
The size of the chunks to create