Skip to content
anavem.com

AI agent

CrewAI

Open-source Python framework for multi-agent workflows. Crews are teams of role-based agents; Flows are event-driven workflows. A separate commercial control plane, CrewAI AMP, adds managed deployment and observability.

Maintainer
crewAI, Inc. (crewAIInc organisation on GitHub)
Licence
MIT (README: "CrewAI is released under the MIT License"; the LICENSE file contains the MIT permission text, Copyright (c) 2025 crewAI, Inc., without a title line)
Last release
1.15.23, released 2026-09-28 (marked Latest on the GitHub releases page)
Last verified Jump to what it can access ↓

What it can access

An agent can take actions, not only answer questions, so what it is allowed to do on your behalf matters most. This is what the listing states, based on the sources below. Anavem does not rate it safe or unsafe: check it against what you plan to use it for.

Permissions it asks for

  • Agents call tools you list per agent; the tools docs list tools for file and document search, web scraping, web search (SerperDev, Exa), GitHub repository search, PostgreSQL search, a code interpreter and image generation
  • Custom tools are written by subclassing BaseTool or with the @tool decorator
  • Needs your model provider API key; the README says agents use the OpenAI API by default and can be pointed at other models, including local ones via Ollama
  • A web search tool such as SerperDevTool needs its own API key (SERPER_API_KEY in the README example)

Data it can reach

Docs state: CrewAI runs in your Python environment, and agents reach only what their tools and model connection expose. Credentials go in a .env file ("Do not hardcode API keys. Use .env." in the installation docs). The README documents anonymous telemetry (its list: CrewAI and Python versions, general OS, number of agents and tasks, the crew process, memory/delegation use, parallel or sequential tasks, the language model, agent roles and tool names) and says users can disable it by setting OTEL_SDK_DISABLED to true. It says no data is collected on prompts, task descriptions, agents' backstories or goals, usage of tools, API calls, responses, processed data, secrets or environment variables, except under the conditions it mentions: when the share_crew feature is enabled, detailed data including task descriptions and agents' backstories or goals is collected. The commercial CrewAI AMP Suite offers on-premise or cloud deployment; the Crew Control Plane is described as having a free part at app.crewai.com. (README, docs.crewai.com, 2026-10-03)

How it is installed or connected

From the repository README and installation guide (https://docs.crewai.com/en/installation), accessed 2026-10-03. The README states Python >=3.10 and <3.14 and uses uv for dependency management.

uv tool install crewai
crewai create crew <name>

The generated project has agents/*.jsonc, crew.jsonc, a .env file, knowledge/, skills/ and tools/ folders. The README says to put the required keys in .env (model provider key, and a Serper.dev key if you use web search), then run crewai install and crewai run from the project directory. The README also offers an older Python/YAML scaffold with crewai create crew <name> --classic.

Limitations

  • Developer framework: you edit agent and crew definition files and manage your own model keys and tool keys
  • The docs page on tools shows a code interpreter tool but states no sandboxing or execution restrictions for it; check this before giving an agent code execution
  • README: CrewAI uses anonymous telemetry; users can disable it by setting OTEL_SDK_DISABLED to true
  • Managed deployment, governance and observability are part of the commercial CrewAI AMP Suite, not the MIT-licensed package; the AMP pricing page was not read for this listing

Not sure what to look for? Read what to check before installing.

Quick answers

Who maintains this AI agent?
crewAI, Inc. (crewAIInc organisation on GitHub).
What can it access?
It asks for: Agents call tools you list per agent; the tools docs list tools for file and document search, web scraping, web search (SerperDev, Exa), GitHub repository search, PostgreSQL search, a code interpreter and image generation, Custom tools are written by subclassing BaseTool or with the @tool decorator, Needs your model provider API key; the README says agents use the OpenAI API by default and can be pointed at other models, including local ones via Ollama, A web search tool such as SerperDevTool needs its own API key (SERPER_API_KEY in the README example). Docs state: CrewAI runs in your Python environment, and agents reach only what their tools and model connection expose. Credentials go in a .env file ("Do not hardcode API keys. Use .env." in the installation docs). The README documents anonymous telemetry (its list: CrewAI and Python versions, general OS, number of agents and tasks, the crew process, memory/delegation use, parallel or sequential tasks, the language model, agent roles and tool names) and says users can disable it by setting OTEL_SDK_DISABLED to true. It says no data is collected on prompts, task descriptions, agents' backstories or goals, usage of tools, API calls, responses, processed data, secrets or environment variables, except under the conditions it mentions: when the share_crew feature is enabled, detailed data including task descriptions and agents' backstories or goals is collected. The commercial CrewAI AMP Suite offers on-premise or cloud deployment; the Crew Control Plane is described as having a free part at app.crewai.com. (README, docs.crewai.com, 2026-10-03). We do not label anything safe or unsafe; read the official sources before you install.
What licence does it use?
MIT (README: "CrewAI is released under the MIT License"; the LICENSE file contains the MIT permission text, Copyright (c) 2025 crewAI, Inc., without a title line). Check the terms if you plan to use it commercially.
What are its limitations?
Developer framework: you edit agent and crew definition files and manage your own model keys and tool keys. The docs page on tools shows a code interpreter tool but states no sandboxing or execution restrictions for it; check this before giving an agent code execution. README: CrewAI uses anonymous telemetry; users can disable it by setting OTEL_SDK_DISABLED to true. Managed deployment, governance and observability are part of the commercial CrewAI AMP Suite, not the MIT-licensed package; the AMP pricing page was not read for this listing.
When was it last released?
1.15.23, released 2026-09-28 (marked Latest on the GitHub releases page). Verified Oct 3, 2026.

Sources

Other listings in the same category.

All AI agents →

AI tools in AI Automation & Agents

Verified tool profiles in the same category.

See category →