CAMEL

CAMEL is a Python framework for developers and researchers building AI agent teams with models running locally through Ollama or through remote providers. Its Ollama integration connects agents to a local model service; choosing OpenAI instead sends model requests to OpenAI and requires an API key. The framework is open source under Apache 2.0.

RolePlaying lets agents work together through assigned roles, while Workforce supports teams with hierarchies and tasks that stretch across many interactions. Agents retain context between steps. The toolkit covers planning, messaging and evaluation, with human participation for workflows that need oversight.

Synthetic data generation is a substantial part of CAMEL's scope. It includes instruction and chain-of-thought pipelines, plus Python, math and physics verifiers. Training workflows connect generated data and interaction logs to fine-tuning or reinforcement learning, with documented examples using Unsloth and Qwen.

Agents can browse websites, execute code and retrieve information from documents or knowledge graphs. MCP support and integrations such as GitHub, Slack and Google Drive connect them to external tools and services, so a local model doesn't make every workflow offline.

For research, CAMEL supports simulated environments and standardized benchmarks for comparing agent behavior. Its ecosystem also provides datasets on Hugging Face covering code, mathematics and scientific subjects.