Web and Network

HTTP, Slack, browser automation, search APIs, and MCP connectivity.

← Prebuilt Tools overview

Tool	Extra	Description
http_request	base	Authenticated HTTP requests and web fetches.
slack	base	Slack Socket Mode and full API access.
slack_send_message	base	Post a message to a Slack channel or thread.
browser	local-chromium-browser / agent-core-browser	Browser automation via LocalChromiumBrowser.browser or AgentCoreBrowser.browser.
rss	rss	Subscribe to and process RSS/Atom feeds.
bright_data	base	Web scraping and search via Bright Data.
tavily_search	base	AI-optimized web search.
tavily_extract	base	Extract clean text from URLs.
tavily_crawl	base	Crawl a site from a base URL.
tavily_map	base	Map site link structure.
exa_search	base	Semantic web search via Exa.
exa_get_contents	base	Fetch full page contents for URLs.
mcp_client	base	Connect to MCP servers at runtime.

---

http_request

Authenticated HTTP requests and web fetches.

from elsai_tools.http_request import http_request

Parameter	Type	Required	Description
method	str	Yes	HTTP method
url	str	Yes	Request URL
headers	dict	No	Request headers
body	str	No	Request body
auth_type	str	No	Bearer, JWT, Basic, Digest, aws_sigv4, etc.
auth_token	str	No	Auth token
basic_auth	dict	No	Basic auth credentials
timeout	int	No	Timeout in seconds

slack

Slack Socket Mode and full API access.

from elsai_tools.slack import slack

Parameter	Type	Required	Description
action	str	Yes	Slack API method name
parameters	Dict[str, Any]	No	Method parameters dict

slack_send_message

Post a message to a Slack channel or thread.

from elsai_tools.slack import slack_send_message

Parameter	Type	Required	Description
channel	str	Yes	Channel ID or name
text	str	Yes	Message text
thread_ts	str	No	Thread timestamp for replies

browser

Browser automation via LocalChromiumBrowser.browser or AgentCoreBrowser.browser.

from elsai_tools.browser import LocalChromiumBrowser

browser = LocalChromiumBrowser()
agent = Agent(tools=[browser.browser])

Extra: local-chromium-browser / agent-core-browser

Parameter	Type	Required	Description
browser_input.action	str	Yes	init_session, navigate, click, close, etc.
browser_input.session_name	str	No	Browser session name
browser_input.url	str	No	URL to navigate
browser_input.selector	str	No	CSS selector

rss

Subscribe to and process RSS/Atom feeds.

from elsai_tools.rss import rss

Extra: rss

Parameter	Type	Required	Description
action	str	Yes	Feed action
url	str	No	Feed URL
feed_id	str	No	Registered feed ID
max_entries	int	No	Max entries to return
include_content	bool	No	Include full content
query	str	No	Filter query

bright_data

Web scraping and search via Bright Data.

from elsai_tools.bright_data import bright_data

Parameter	Type	Required	Description
action	str	Yes	Bright Data action
**options	Any	No	Action-specific options

tavily_search

AI-optimized web search.

from elsai_tools.tavily import tavily_search

Parameter	Type	Required	Description
query	str	Yes	Search query
search_depth	Literal['basic', 'advanced']	No	basic or advanced
topic	Literal['general', 'news']	No	general or news
max_results	int	No	Max results (0–20)
include_domains	Any	No	Domains to include
exclude_domains	Any	No	Domains to exclude

tavily_extract

Extract clean text from URLs.

from elsai_tools.tavily import tavily_extract

Parameter	Type	Required	Description
urls	Union[str, List[str]]	Yes	URL or list of URLs
extract_depth	Literal['basic', 'advanced']	No	basic or advanced
format	Literal['markdown', 'text']	No	markdown or text
include_images	bool	No	Include images

tavily_crawl

Crawl a site from a base URL.

from elsai_tools.tavily import tavily_crawl

Parameter	Type	Required	Description
url	str	Yes	Starting URL
max_depth	int	No	Crawl depth
limit	int	No	Max pages
instructions	str	No	Crawl instructions
extract_depth	Literal['basic', 'advanced']	No	Extraction depth

tavily_map

Map site link structure.

from elsai_tools.tavily import tavily_map

Parameter	Type	Required	Description
url	str	Yes	Starting URL
max_depth	int	No	Map depth
limit	int	No	Max URLs
select_paths	List[str]	No	Paths to include
exclude_paths	List[str]	No	Paths to exclude

exa_search

Semantic web search via Exa.

from elsai_tools.exa import exa_search

Parameter	Type	Required	Description
query	str	Yes	Search query
type	Any	No	auto, instant, fast, or deep
num_results	Any	No	Number of results
include_domains	Any	No	Domains to include
exclude_domains	Any	No	Domains to exclude

exa_get_contents

Fetch full page contents for URLs.

from elsai_tools.exa import exa_get_contents

Parameter	Type	Required	Description
urls	List[str]	Yes	List of URLs
text	Union[bool, Dict[str, Any]]	No	Include full text
highlights	Union[bool, Dict[str, Any]]	No	Include highlights
summary	Dict[str, Any]	No	Generate summary
livecrawl	Literal['never', 'fallback', 'always', 'preferred']	No	Live crawl mode

mcp_client

Connect to MCP servers at runtime.

from elsai_tools.mcp_client import mcp_client

Parameter	Type	Required	Description
action	str	Yes	MCP action
server_config	Dict[str, Any]	No	Server connection config
connection_id	str	No	Active connection ID
tool_name	str	No	Tool to invoke
tool_args	Dict[str, Any]	No	Tool arguments

mcp_client vs SDK MCPClient

mcp_client is an agent tool — the model connects to MCP servers during a run. The SDK's MCPClient is used by you at agent setup to register MCP tools statically.