AgentLens

AgentLens is a free tool for developers to fix problems in AI agents. It lets you record, watch again, and study how agents work on your own computer. This helps you find and fix issues without spending extra money on API calls. It is great for people who want to fix problems quickly and easily on their own computer.

Benefits
\nAgentLens has many good points that make it helpful for developers.

\n
• Offline Debugging: Record and watch agent runs without making extra API calls. This means you can fix problems on your own computer without extra costs.
\n
• Failure Analysis: Find common problems like timeouts or empty outputs. This helps you spot and fix issues quickly.
\n
• Cost Tracking: Keep an eye on how much you are using and how much it costs. This helps you spend money wisely and avoid surprises.
\n
• Local-First Workflow: Everything runs on your computer. This makes debugging secure and efficient with no need for outside help.
\n

Use Cases
\nAgentLens can be used in many ways.

\n
• Development Iteration: Fix agents on your own computer without calling APIs over and over. This speeds up your work.
\n
• Cost Optimization: Find expensive or wasteful patterns in how agents work. This helps you save on API costs.
\n
• Regression Testing: Check if agents work right after changes to make sure they are stable and reliable.
\n
• Educational Purposes: Study how agents make decisions. This helps you understand how your AI agents work better.
\n

Additional Information
\nAgentLens is part of a bigger set of tools for making AI agents. It has simple ways to build and check agent structures. The tool works with Python function decorators, context managers, LangChain, and the OpenAI Python client. Support for CrewAI and other frameworks is coming soon.

AgentLens is free to use under the AGPLv3 license, and the data is free to share under the CC-BY-SA 4.0 license. Everyone can help make it better, and you can find more details in the CONTRIBUTING.md file. For more information, visit the AgentLens GitHub page at https://github.com/auriel-ai/agentlens.