Strands Agents SDK integration

Copy for LLM

The Strands Agents SDK is an open-source Python SDK by AWS for building AI agents.

It follows a model-tools-prompt pattern: the SDK sends the prompt to the model, executes any tool calls the model requests, and feeds the results back until the task is complete.

The strands-apify package provides 18 tools grouped into three tool sets, giving your agent scraping, crawling, search, and social media capabilities through the Apify platform.

Help keep this page up to date

This integration uses a third-party service. If you find outdated content, please submit an issue on GitHub.

Prerequisites

• Python 3.11 or newer
• An Apify account and Apify API token from the Integrations page in Apify Console
• At least one configured model provider, such as Amazon Bedrock, OpenAI, Anthropic, Google Gemini, Ollama, LiteLLM, or LMStudio

Installation

Install the Strands SDK and the Apify tools:

pip install strands-agents strands-apify

Export your Apify API token:

export APIFY_TOKEN=your_apify_api_token_here

Keep your token out of source code

Use a .env file or your secrets manager in production. Never commit APIFY_TOKEN to a repository.

Environment variables

The package reads the following environment variables:

Variable	Description	Required
APIFY_TOKEN	Apify API token used to authorize Actor runs and dataset reads.	Yes
STRANDS_APIFY_QUIET	Set to 1 to suppress the diagnostic panel printed on each tool call. Auto-suppressed in non-interactive environments.	No
STRANDS_APIFY_MAX_TIMEOUT_SECS	Ceiling for timeout_secs. Default 600.	No
STRANDS_APIFY_MAX_ITEMS	Ceiling for limit, dataset_items_limit, results_limit, search_limit, max_reviews, and max_pages. Default 1000. LinkedIn profile posts stays at its own 100 unless this is set lower.	No
STRANDS_APIFY_MAX_CRAWL_DEPTH	Ceiling for max_depth. Default 5.	No
STRANDS_APIFY_MAX_MEMORY_MBYTES	Ceiling for memory_mbytes. Default and maximum 32768; it can only be lowered.	No

Limits on agent-supplied values

Tools cap numeric arguments at the ceilings in the table before calling Apify. memory_mbytes is rounded up to a power of 2 and is at least 128. For a longer run, raise STRANDS_APIFY_MAX_TIMEOUT_SECS. To read past the item ceiling, call apify_get_dataset_items with offset.

Configure your model provider

Install the matching extra, set the provider's API key, and create a model instance to pass to your agent.

Amazon Bedrock

Bedrock is the default provider and ships with the base strands-agents install. Configure your AWS credentials with aws configure or environment variables before running your agent. See the Bedrock provider docs for the full configuration reference.

from strands.models import BedrockModel

model = BedrockModel(model_id="global.anthropic.claude-sonnet-4-6", region_name="us-west-2")

OpenAI

See the OpenAI provider docs for the full configuration reference.

pip install 'strands-agents[openai]'
export OPENAI_API_KEY=your_openai_key

import os
from strands.models.openai import OpenAIModel

model = OpenAIModel(
    client_args={"api_key": os.getenv("OPENAI_API_KEY")},
    model_id="gpt-4.1",
)

Anthropic

See the Anthropic provider docs for the full configuration reference.

pip install 'strands-agents[anthropic]'
export ANTHROPIC_API_KEY=your_anthropic_key

import os
from strands.models.anthropic import AnthropicModel

model = AnthropicModel(
    client_args={"api_key": os.getenv("ANTHROPIC_API_KEY")},
    model_id="claude-sonnet-4-6",
    max_tokens=1024,
)

Ollama (local)

No API key required. See the Ollama provider docs for the full configuration reference.

pip install 'strands-agents[ollama]'
ollama pull qwen3.5

from strands.models.ollama import OllamaModel

model = OllamaModel(host="http://localhost:11434", model_id="qwen3.5")

Other providers

Strands also supports Google Gemini, LiteLLM, and LMStudio. See the Strands model providers documentation for the full list.

Quick start

This example fetches a single web page and summarizes it:

from strands import Agent
from strands_apify import apify_web_fetch

# `model` comes from the "Configure your model provider" section above.
agent = Agent(model=model, tools=[apify_web_fetch])

agent("Fetch https://docs.apify.com and summarize the page in three bullet points.")

The agent calls apify_web_fetch, returns markdown to the model, and the model produces the summary.

Choose the right tool set

strands-apify provides 18 tools split across three tool sets:

Tool set	Tools	Use case
APIFY_CORE_TOOLS	6	Run any Actor or saved task, fetch dataset items, fetch single web pages
APIFY_SEARCH_TOOLS	5	Google Search, Google Maps, YouTube, multi-page website crawling, e-commerce
APIFY_SOCIAL_TOOLS	7	Instagram, LinkedIn, Twitter/X, TikTok, Facebook

LLMs select tools based on their names and descriptions. More tools means a larger decision space, which can lead to wrong tool selection, slower responses, and higher token usage. Register only the tools your agent needs.

Three ways to import tools:

1. Import a whole tool set when your agent needs the full category:

   from strands import Agent
   from strands_apify import APIFY_CORE_TOOLS
   
   agent = Agent(model=model, tools=APIFY_CORE_TOOLS)

2. Import individual tools for tighter control:

   from strands_apify import apify_web_fetch, apify_google_search_scraper
   
   agent = Agent(model=model, tools=[apify_web_fetch, apify_google_search_scraper])

3. Mix tool sets with individual tools:

   from strands_apify import APIFY_CORE_TOOLS, apify_twitter_scraper
   
   agent = Agent(model=model, tools=[*APIFY_CORE_TOOLS, apify_twitter_scraper])

Avoid APIFY_ALL_TOOLS in production

APIFY_ALL_TOOLS bundles every tool from all three tool sets. Registering 18 tools at once degrades model accuracy. Use a focused tool set in production.

Examples

The web research example builds its own OpenAI model. The other examples reuse the model from Configure your model provider.

Web research agent

This agent searches Google, fetches the top result, and summarizes it:

import os
from strands import Agent
from strands.models.openai import OpenAIModel
from strands_apify import apify_google_search_scraper, apify_web_fetch

model = OpenAIModel(
    client_args={"api_key": os.getenv("OPENAI_API_KEY")},
    model_id="gpt-4.1",
)

agent = Agent(model=model, tools=[apify_google_search_scraper, apify_web_fetch])

agent(
    "Search Google for 'Strands Agents SDK AWS' and return the top 3 results. "
    "Then fetch the first result and summarize the page content in 3 bullet points."
)

Social media monitoring agent

This agent fetches recent tweets and Instagram posts on a topic:

from strands import Agent
from strands_apify import apify_twitter_scraper, apify_instagram_scraper

agent = Agent(model=model, tools=[apify_twitter_scraper, apify_instagram_scraper])

agent("Find the latest 10 tweets mentioning #WebScraping and summarize the main topics.")

Run any Actor from Apify Store

The prebuilt tools cover the most common use cases, but Apify Store has thousands of Actors. Use apify_run_actor_and_get_dataset to give your agent access to any of them.

Unlike prebuilt tools (like apify_google_search_scraper) that expose typed parameters, apify_run_actor_and_get_dataset requires you to provide the Actor's input JSON directly. Your prompt must include the correct input shape.

1. Find the Actor on Apify Store.
2. Check the Input tab for the expected JSON schema.
3. Include the Actor ID and the input JSON in your agent prompt.

from strands import Agent
from strands_apify import apify_run_actor_and_get_dataset

agent = Agent(model=model, tools=[apify_run_actor_and_get_dataset])

agent(
    "Run the Apify Actor 'apify/rag-web-browser' with input "
    '{"query": "latest AI safety research", "maxResults": 5} '
    "and summarize the results."
)

For long-running Actors, such as large crawls or datasets with thousands of items, set timeout_secs in your prompt. See Limits on agent-supplied values for the default ceiling and how to raise it. For example:

agent(
    "Run 'apify/website-content-crawler' with input "
    '{"startUrls": [{"url": "https://docs.example.com"}], "maxCrawlPages": 100} '
    "with timeout_secs=600 and summarize the crawled pages."
)

Include Actor input fields in your prompt

The agent passes run_input as a JSON object to the Actor. For Actors with complex input schemas, spell out the required fields directly in your prompt so the model knows exactly what to provide.

Tool reference

This section lists every tool in strands-apify with its parameters, default values, and an example prompt you can use directly in your agent. Numeric parameters such as timeout_secs, memory_mbytes, and item limits are capped as described in Limits on agent-supplied values.

Core tools

The following tools are included in the APIFY_CORE_TOOLS tool set. They cover running Actors, fetching dataset items, fetching single web pages, and executing saved tasks.

apify_web_fetch

Fetch a single web page and return its content as markdown, HTML, plain text, links, or raw content. Uses Web Fetch, which answers in one real-time request and handles bot protection on its own. No Actor input schema needed. For multi-page crawls, use apify_website_content_crawler.

Parameter	Type	Default	Description
url	str	required	The URL to fetch (must include http:// or https://)
formats	list[str]	["markdown"]	Any of "markdown", "html", "text", "links", and "raw" (original bytes, base64-encoded for binary content). Use ["raw"] for images, archives, and other binary files.
headers	dict	None	HTTP headers to send to the target URL (e.g. {"Accept-Language": "fr-FR"})
wait_ms	int	None	Milliseconds to wait after the page loads, from 1 to 60000. Renders the page in a real browser so content loaded later by JavaScript is included, but makes the request slower.
timeout_secs	int	180	Maximum wait time in seconds

The result includes url, final_url (after redirects), title, http_status (the target page's own status, e.g. 404 for a missing page), and one key per requested format.

Example prompt:

Fetch https://docs.apify.com/academy and summarize the page.

apify_run_actor

Run any Actor from Apify Store and return run metadata only (run ID, status, dataset ID). Use this when you need the run metadata but will fetch results separately with apify_get_dataset_items.

Parameter	Type	Default	Description
actor_id	str	required	Actor identifier in "username/actor-name" format
run_input	dict	None	JSON-serializable input matching the Actor's input schema
timeout_secs	int	300	Maximum wait time in seconds
memory_mbytes	int	None	Memory allocation in MB (uses Actor default if not set)
build	str	None	Build tag or number to pin a specific Actor version

To find the input schema, open the Actor's page on Apify Store and check the Input tab. The run_input JSON object must match that schema.

Example prompt:

Run the Actor 'apify/website-content-crawler' with input {"startUrls": [{"url": "https://example.com"}], "maxCrawlPages": 5} and return the run metadata.

apify_run_actor_and_get_dataset

Run any Actor and fetch its dataset results in a single call. Provides access to thousands of Actors in Apify Store, but requires you to provide the Actor's input JSON.

Parameter	Type	Default	Description
actor_id	str	required	Actor identifier in "username/actor-name" format
run_input	dict	None	JSON-serializable input matching the Actor's input schema
timeout_secs	int	300	Maximum wait time in seconds
memory_mbytes	int	None	Memory allocation in MB (uses Actor default if not set)
build	str	None	Build tag or number to pin a specific Actor version
dataset_items_limit	int	100	Maximum dataset items to return
dataset_items_offset	int	0	Items to skip (for pagination)

Example prompt:

Run the Actor 'apify/rag-web-browser' with input {"query": "latest AI safety research", "maxResults": 5} and summarize the results.

The tool doesn't validate run_input against the Actor's input schema. A mismatch fails on the Actor run.

How to use this tool effectively

1. Browse Apify Store and find the Actor you need.
2. Open the Actor's page and check the Input tab for the JSON schema.
3. In your agent prompt, provide the Actor ID and the input as a JSON object.

apify_get_dataset_items

Fetch items from an existing Apify dataset. Use this after apify_run_actor to get results, or to paginate through large datasets.

Parameter	Type	Default	Description
dataset_id	str	required	The dataset ID (returned by apify_run_actor)
limit	int	100	Maximum items to return
offset	int	0	Items to skip (for pagination)

Example prompt:

Fetch the first 50 items from dataset ID 'abc123' and summarize the key findings.

apify_run_task

Run a saved Actor task with optional input overrides. Tasks are pre-configured Actor runs saved in Apify Console.

Parameter	Type	Default	Description
task_id	str	required	Task identifier in "username/task-name" format or a task ID
task_input	dict	None	Optional input fields to override the task's saved defaults
timeout_secs	int	300	Maximum wait time in seconds
memory_mbytes	int	None	Memory allocation in MB (uses task default if not set)

Example prompt:

Run my saved task 'john/daily-news-scrape' and return the run status.

apify_run_task_and_get_dataset

Run a saved task and fetch its dataset results in one call.

Parameter	Type	Default	Description
task_id	str	required	Task identifier in "username/task-name" format or a task ID
task_input	dict	None	Optional input fields to override the task's saved defaults
timeout_secs	int	300	Maximum wait time in seconds
memory_mbytes	int	None	Memory allocation in MB (uses task default if not set)
dataset_items_limit	int	100	Maximum dataset items to return
dataset_items_offset	int	0	Items to skip (for pagination)

Example prompt:

Run my task 'john/daily-news-scrape' and summarize the top 10 results.

---

Search and crawling tools

The following tools are included in the APIFY_SEARCH_TOOLS tool set. They cover Google Search, Google Maps, YouTube, multi-page website crawling, and e-commerce scraping.

apify_google_search_scraper

Search Google and return structured results (organic links, ads, and People Also Ask). Uses Google Search Scraper.

Parameter	Type	Default	Description
search_query	str	required	Google search query. Supports operators like site:, "exact phrase", OR
results_limit	int	10	Maximum results to return (multiples of 10 trigger extra pages)
country_code	str	None	Two-letter country code for localized results (e.g. "us", "de")
language_code	str	None	Two-letter language code (e.g. "en", "es")
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Search Google for 'best python web frameworks 2025' and return the top 5 results.

apify_google_places_scraper

Search Google Maps for businesses and places, optionally with reviews. Uses Google Maps Scraper.

Parameter	Type	Default	Description
search_query	str	required	Google Maps search query (e.g. "restaurants in Prague")
results_limit	int	20	Maximum places to return
language	str	None	Language for results (e.g. "en", "de")
include_reviews	bool	False	Whether to include user reviews
max_reviews	int	5	Reviews per place (only used when include_reviews=True)
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Find the top-rated Italian restaurants in Berlin with reviews.

apify_youtube_scraper

Scrape YouTube videos, channels, or search results. Uses YouTube Scraper. Provide at least one of search_query or urls.

Parameter	Type	Default	Description
search_query	str	None	YouTube search query (e.g. "python tutorial")
urls	list[str]	None	Specific YouTube video or channel URLs to scrape
results_limit	int	20	Maximum results to return
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Search YouTube for 'AWS re:Invent 2025 keynote' and return the top 5 videos with their view counts.

apify_website_content_crawler

Crawl a website and extract content from multiple pages as markdown. Uses Website Content Crawler. This is the multi-page version. For single pages, use apify_web_fetch.

Parameter	Type	Default	Description
start_url	str	required	Starting URL to crawl from
max_pages	int	10	Maximum number of pages to crawl
max_depth	int	2	Maximum link depth from the start URL
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Crawl https://docs.example.com up to 20 pages and summarize the documentation structure.

apify_ecommerce_scraper

Scrape product data from e-commerce sites (Amazon, eBay, Walmart, and others). Uses E-commerce Scraping Tool. The Actor auto-detects the platform.

Parameter	Type	Default	Description
url	str	required	Product or listing page URL
url_type	str	"product"	"product" for a single product page, "listing" for a category/search results page
results_limit	int	20	Maximum products to return (relevant for listings)
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Scrape the product details from https://www.amazon.com/dp/B0EXAMPLE and return the title, price, and rating.

---

Social media tools

The following tools are included in the APIFY_SOCIAL_TOOLS tool set. They cover Instagram, LinkedIn, Twitter/X, TikTok, and Facebook.

apify_instagram_scraper

Scrape Instagram profiles, posts, reels, or hashtags. Uses Instagram Scraper. Provide at least one of search_query or urls.

Parameter	Type	Default	Description
search_query	str	None	Username, hashtag, or keyword. If it looks like an Instagram URL, it's treated as a direct URL.
urls	list[str]	None	Direct Instagram URLs to scrape (profiles, posts, reels)
results_type	str	"posts"	What to scrape: "posts", "comments", or "details" (profile metadata)
results_limit	int	20	Maximum items per URL or search hit
search_type	str	"hashtag"	Search mode: "hashtag", "user", or "place"
search_limit	int	10	How many search results to process (each yields up to results_limit items)
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Scrape the latest 15 posts from the hashtag #webdevelopment on Instagram.

apify_linkedin_profile_posts

Scrape recent posts from a LinkedIn profile. Uses LinkedIn Profile Posts.

Parameter	Type	Default	Description
profile_url	str	required	LinkedIn profile URL (e.g. "https://www.linkedin.com/in/username") or bare username
results_limit	int	20	Maximum posts to return (capped at 100)
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Get the last 10 posts from https://www.linkedin.com/in/satyanadella and summarize the main topics.

apify_linkedin_profile_search

Search for LinkedIn profiles by keywords with optional filters. Uses LinkedIn Profile Search.

Parameter	Type	Default	Description
search_query	str	required	Keywords like job titles, skills, or names (e.g. "software engineer")
results_limit	int	20	Maximum profiles to return
locations	list[str]	None	Filter by locations (e.g. ["San Francisco", "New York"])
current_job_titles	list[str]	None	Filter by current job title (e.g. ["CTO", "VP Engineering"])
profile_scraper_mode	str	"Short"	"Short" for basic data, "Full" for complete profile details (slower)
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Find 10 machine learning engineers in London and return their profile summaries.

apify_linkedin_profile_detail

Get full details from a single LinkedIn profile (experience, education, and skills). Uses LinkedIn Profile Detail. No LinkedIn account or cookies required.

Parameter	Type	Default	Description
profile_url	str	required	LinkedIn profile URL or bare username
include_email	bool	False	Include email if publicly available
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Get the full profile details for https://www.linkedin.com/in/example-user including their work experience and education.

apify_twitter_scraper

Scrape tweets from Twitter/X by search, handles, or URLs. Uses Twitter Scraper Lite. Supports Twitter advanced search operators. Provide at least one of search_query, urls, or twitter_handles.

Parameter	Type	Default	Description
search_query	str	None	Search query with support for operators like from:user, #hashtag, min_faves:N, since:YYYY-MM-DD
urls	list[str]	None	Specific tweet, profile, or list URLs
twitter_handles	list[str]	None	Handles without @ (e.g. ["NASA", "WHO"])
results_limit	int	20	Maximum tweets to return
sort	str	"Latest"	"Latest" (chronological) or "Top" (most popular)
tweet_language	str	None	ISO 639-1 language code (e.g. "en", "es")
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Find the latest 20 tweets mentioning #AI from the past week, sorted by most popular.

apify_tiktok_scraper

Scrape TikTok videos by search, hashtag, profile, or direct URL. Uses TikTok Scraper. Provide at least one of search_query, hashtags, profiles, or urls.

Parameter	Type	Default	Description
search_query	str	None	Keyword to search TikTok
hashtags	list[str]	None	Hashtags without # (e.g. ["fyp", "cooking"])
profiles	list[str]	None	TikTok usernames to scrape videos from
urls	list[str]	None	Specific TikTok post URLs
results_limit	int	20	Maximum videos per source
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Scrape the top 10 TikTok videos for the hashtag 'programming' and return their view counts.

apify_facebook_posts_scraper

Scrape posts from a Facebook page or profile. Uses Facebook Posts Scraper.

Parameter	Type	Default	Description
page_url	str	required	Facebook page or profile URL
results_limit	int	20	Maximum posts to return
only_posts_newer_than	str	None	Date filter: "2024-01-01", "1 week ago", "3 months ago"
timeout_secs	int	300	Maximum wait time in seconds

Example prompt:

Get the last 15 posts from https://www.facebook.com/apifytech newer than 1 month ago.

---

Deprecated tools

The following tools aren't part of any tool set and will be removed in a future release. Existing code that imports them directly keeps working.

apify_scrape_url

apify_scrape_url is deprecated. Use apify_web_fetch instead. Calling it emits a FutureWarning.

Parameter	Type	Default	Description
url	str	required	The URL to scrape (must include http:// or https://)
timeout_secs	int	120	Maximum wait time in seconds
crawler_type	str	"cheerio"	Engine: "cheerio" (fast, no JS), "playwright:adaptive" (renders JS when needed), "playwright:firefox" (full JS, best anti-bot bypass)

Production environment best practices

• Increase timeout_secs when crawling large sites or scraping high-volume datasets. See Limits on agent-supplied values for the ceiling.
• Lower the STRANDS_APIFY_MAX_* ceilings to limit how much a single agent tool call can run or return.
• Paginate datasets with apify_get_dataset_items using limit and offset instead of fetching everything in one call.
• Pin Actor versions by passing the build parameter (e.g. build="1.2.3" or build="beta") to apify_run_actor / apify_run_actor_and_get_dataset.
• Monitor Actor costs in Apify Console. Agentic loops can call Actors multiple times per user request. Set a billing limit to control spending.
• Register only the tools your agent needs. For example, if you only search Google and scrape one URL, register exactly those two tools instead of APIFY_SEARCH_TOOLS or APIFY_ALL_TOOLS.
• Set STRANDS_APIFY_QUIET=1 to hide the diagnostic panel in an interactive shell. Non-interactive environments (CI, Docker, web services, and Lambda) hide it on their own.

Troubleshooting

Problem	Cause	Fix
APIFY_TOKEN environment variable is not set	Token not configured	Run export APIFY_TOKEN=your_apify_token before executing your script
apify-client package is required	Missing dependency	Run pip install strands-apify
Invalid URL or urls[N]: Invalid URL scheme '...'	URL is not well-formed http(s)	Confirm the URL has an http:// or https:// scheme and a valid host. The index in urls[N] points to the offending entry.
Actor run finishes with status FAILED	Invalid input or Actor error	Check the statusMessage in the error text, or check the full run logs in Apify Console.
Actor run finishes with status TIMED-OUT	Timeout too short for the workload	Increase the timeout_secs parameter. For runs longer than 600 seconds, also raise STRANDS_APIFY_MAX_TIMEOUT_SECS.
Agent selects the wrong tool	Too many tools registered or ambiguous prompt	Reduce the number of registered tools, add more context to the prompt, or use a more capable model
Empty results from social media tools	Private or geo-restricted profile	Verify the profile is public. Test with a known public account first.

Resources

• Strands Agents SDK documentation
• Strands Agents SDK GitHub repository
• strands-apify on PyPI
• strands-apify GitHub repository
• Apify Store
• Apify API reference