CAMEL

CAMEL is an open-source Python framework for developers and researchers building systems where AI agents work together. Its focus is on agent roles, communication, and behavior across extended tasks, with applications in synthetic training data, task automation, and simulated societies. It uses the Apache 2.0 license.

Role-playing agents can carry out conversations with assigned responsibilities, while Workforce models teams with roles and hierarchies for longer tasks. Stateful memory keeps historical context available across interactions. The framework also includes task planning, messaging, evaluation, and observability tools for studying how those systems perform.

Data generation is a central use case. CAMEL supports instruction and chain-of-thought datasets, including pipelines that improve generated reasoning through verification. It connects interaction logs to reinforcement learning and fine-tuning workflows, and includes Python, math, and physics verifiers.

For agents that need information or actions beyond a conversation, CAMEL includes browser tools, code execution, document processing, and retrieval, including graph-based retrieval. Named integrations include GitHub, Slack, Google Drive, Notion, and MCP. Human-in-the-loop tools let people participate in agent workflows.

Researchers can use simulated environments and standardized benchmarks to compare agent performance. The ecosystem also provides datasets hosted on Hugging Face for code, math, physics, chemistry, and biology.