GitHub Repository

Self-hosted AI agent observability with tool-call tracing and decision tree visualization

3 starsPython

AgentLens – Open-source observability for AI agents

Name: AgentLens – Open-source observability for AI agents
Availability: InStock
Author: tranhoangtu

by tranhoangtu·Mar 1, 2026·2 points·2 comments

Visit Project View on HN

AI Analysis

●●●BangerSolve My ProblemBig Brain

Agent-native observability with DAG topology when LangSmith and Langfuse miss multi-agent flows.

Strengths

•Genuinely solves a pain point: LangSmith is paid/cloud-only, Langfuse doesn't understand agent topology or handoffs
•Time-travel replay + span comparison is clever debugging UX not found in competitors
•Self-hosted, MIT, 231 tests with 100% coverage; works with all major agent frameworks

Weaknesses

•Requires self-hosting and manual instrumentation; early-stage (231 tests suggests smaller surface area)
•Adoption depends on LangChain/CrewAI ecosystem growth; niche audience relative to general observability tools

Post Description

Hi HN,

I built AgentLens because debugging multi-agent systems is painful. LangSmith is cloud-only and paid. Langfuse tracks LLM calls but doesn't understand agent topology — tool calls, handoffs, decision trees.

AgentLens is a self-hosted observability platform built specifically for AI agents:

- *Topology graph* — see your agent's tool calls, LLM calls, and sub-agent spawns as an interactive DAG - *Time-travel replay* — step through an agent run frame-by-frame with a scrubber timeline - *Trace comparison* — side-by-side diff of two runs with color-coded span matching - *Cost tracking* — 27 models priced (GPT-4.1, Claude 4, Gemini 2.0, etc.) - *Live streaming* — watch spans appear in real-time via SSE - *Alerting* — anomaly detection for cost spikes, error rates, latency - *OTel ingestion* — accepts OTLP HTTP JSON, so any OTel-instrumented app works

Works with LangChain, CrewAI, AutoGen, LlamaIndex, and Google ADK.

Tech: React 19 + FastAPI + SQLite/PostgreSQL. MIT licensed. 231 tests, 100% coverage.

docker run -p 3000:3000 tranhoangtu/agentlens-observe:0.6.0 pip install agentlens-observe

Demo GIF and screenshots in the README.

GitHub: https://github.com/tranhoangtu-it/agentlens-observe Docs: https://agentlens-observe.pages.dev

I'd love feedback on the trace visualization approach and what features matter most for your agent debugging workflow.