dunetrace
Active·★ 68·NOASSERTION·Updated 2026-09-20
Dunetrace monitors every run of AI agents in real-time, detecting structural failures like tool loops, retry storms, and context bloat within 15 seconds of completion. It fires alerts via Slack or webhook, provides plain-English explanations, and offers one-click fixes through Langfuse or GitHub PRs.
#agent-observability#ai#ai-agents#ai-tools#analytics#autogen#crewai#langchain
01
Features
01Real-time monitoring with 15-second alert latency
0217 automatic structural detectors (no configuration needed)
03Diagnose with plain-English explanations and LLM-powered root-cause analysis (via Langfuse)
04One-click fix application (prompt version or GitHub PR)
05Policies for mid-run guardrails (cost cap, tool call limit)
02
Compatibility
Python
SDK (Python)
Verified via docs
Node.js
SDK (Node.js)
Verified via docs
LangChain
LangChain/LangGraph
Verified via docs
Langfuse
Langfuse Deep Analysis
Verified via docs
MCP
MCP Server
Verified via docs
03
Quick start
1
$ git clone https://github.com/dunetrace/dunetrace
2
$ cd dunetrace
3
$ cp .env.example .env
4
$ docker compose build
5
$ docker compose up -d
04
Use cases
↳Monitor production AI agents for silent failures like tool loops and cost spikes
↳Automatically detect and alert on agent behavioral anomalies
↳Quickly diagnose and fix agent issues with one-click actions
05
Alternatives
rustyhand★ 20
Agent OS in Rust — one binary, 37 agents, 26 LLM providers, 37 channels (Telegram/Discord/Slack...), MCP server, A2A protocol, 120+ API endpoints, web dashboard
Octopoda-OS★ 486
The open-source memory operating system for AI agents. Persistent memory, semantic search, loop detection, agent messaging, crash recovery, and real-time observability.
Comments
Log in to leave a comment
No comments yet. Be the first!