AgenticLens

AgenticLens is a browser-based visualization tool that transforms JSONL exports from AI agent frameworks into interactive debugging workspaces. It takes the often-opaque log files generated by tools like the Claude Agent SDK and creates clear, navigable visual representations of agent workflows. You can upload your JSONL files directly in the browser and immediately see your agent's execution path through multiple synchronized views.

The tool stands out by offering four complementary perspectives on the same data: a flow canvas showing step connections, a hierarchical tree view for nested operations, a chronological timeline with performance metrics, and a replay mode for step-by-step execution. What makes AgenticLens particularly useful is its client-side processing—your sensitive agent data never leaves your browser, making it ideal for teams working with proprietary AI systems. It automatically identifies and flags performance bottlenecks with visual badges on slow or high-token steps.

This tool is most valuable for developers building AI agents who need to understand why their systems behave certain ways. It helps debug complex reasoning chains, optimize token usage and latency, and compare different agent runs. Whether you're troubleshooting a failed tool call or analyzing why an agent took a particular path, AgenticLens provides the visual context that raw JSONL files lack.