selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

TraceRoot logo

Self-host TraceRoot

v1.3.3

Open-source AI agent observability and improvement platform

ReleaseOpen CoreOther
Updated11 hours ago
LanguageTypescript
Setup difficultyAdvanced
ThemeDark mode
Stars
1 084
Ports
3000 (Web UI (Next.js frontend)), 8000 (REST API (FastAPI backend)), 8100 (AI Agent Service)
Requires
ClickHouse, MinIO (S3-compatible), PostgreSQL, Redis
Install with Docker Compose
Research agent trace showing nested model and tool calls alongside inputs and outputsDetector results listing evaluated traces, judge outputs, summaries, and linked signalsSignal analysis with root cause, recommended changes, frequency over time, and affected tracesEvaluation run comparison showing outputs and scorer results against a baseline

What is TraceRoot?

TraceRoot helps developers find patterns in production AI agent failures and ship verified fixes. It captures model and tool calls with their inputs, outputs, latency, and cost; runs detectors with LLM judges to evaluate behavior; clusters judge outputs into recurring signals investigated by background coding agents in sandboxes; and runs offline evaluations to compare versions and catch regressions. A command-line interface gives coding agents read/write access to traces, detectors, datasets, evaluations, dashboards, and alerts. Native Python and TypeScript SDKs plus many integrations cover frameworks like LangGraph, CrewAI, AutoGen, and Vercel AI SDK, plus model providers including OpenAI, Anthropic, Gemini, and Mistral. Self-hosting uses Docker Compose with PostgreSQL, ClickHouse, Redis, and MinIO. Core code is Apache 2.0; enterprise features need a separate license.

Should you self-host TraceRoot?

TraceRoot scores 58/100 on selfhost.directory (how scoring works) and ranks #11 of 51 in Monitoring & observability. It has 1 084 GitHub stars.

It shipped 10 stable releases in the past year. The latest, v1.3.3, came out on 27 October 2026. New versions land roughly every 2 days.

Setup is advanced: plan for several services and some ongoing maintenance. It is Other-licensed and written mostly in Typescript.

Its closest alternative on selfhost.directory is GoAccess (75/100, 21 092 stars). It scores higher overall, so compare the two before you pick. See TraceRoot vs GoAccess side by side.

Features

  • 40+ Framework Integrations
  • BYOK (Bring Your Own Keys)
  • Coding Agent CLI
  • Dashboard & Alerting
  • Distributed Tracing
  • Enterprise SSO & RBACRequires enterprise license key (TRACEROOT_EE_LICENSE_KEY)
  • LLM-based Detection
  • Model Provider Integrations
  • Offline Evaluations
  • OpenTelemetry compatibility
  • Python SDK
  • Sandbox-based Code Investigation
  • Signal Clustering
  • TypeScript SDK

TraceRoot README

How to install TraceRoot

Install TraceRoot with Docker Compose

Git, Docker Compose, Make

git clone https://github.com/traceroot-ai/traceroot.git cd traceroot cp .env.example .env make prod-lite

Runs full stack in containers (PostgreSQL, ClickHouse, MinIO, Redis, web, rest, worker, billing, detector, agent). Best for trying TraceRoot locally or production deployment.

TraceRoot releases & version history

FAQ

Can I start for free?

Yes, TraceRoot Cloud has a free plan with no credit card required. Bring your own model API keys or use hosted models.

Is TraceRoot open source? Can I self-host it?

Yes, TraceRoot is open source and can be self-hosted with Docker. Code outside ee directories is Apache 2.0 licensed; enterprise directories use a separate Enterprise License.

Does TraceRoot work with my agent stack?

TraceRoot offers Python and JavaScript/TypeScript SDKs, OpenTelemetry-compatible tracing, and integrations with frameworks such as LangGraph, Vercel AI SDK, and CrewAI.

Can I use TraceRoot with my coding agent?

Yes, the TraceRoot CLI gives coding agents access to production traces, detectors, datasets, evals, dashboards, and alerts. Install with npm install -g traceroot-cli and run traceroot login.

How do detectors and evals work together?

Detectors run online evaluations on production traces to identify failures. Patterns across judge outputs surface as improvement signals. Offline agent evals then test changes against datasets and compare results with a baseline to verify fixes and catch regressions.

How does the improvement loop work?

TraceRoot automatically detects failures in production agent traffic, clusters judge outputs into improvement signals, lets coding agents investigate and make changes in a local worktree, then verifies fixes with offline evals against datasets.

What is TraceRoot?

TraceRoot is an open-source AI agent improvement platform that helps find patterns in production agent issues and ship verified fixes using coding agents.

More in Monitoring & observability

Rate this project

Click to rate — no account needed.

0 comments

Posting as guest
Loading comments…