# truera/trulens

Evaluation and Tracking for LLM Experiments and AI Agents

Repository: https://github.com/truera/trulens
Canonical: https://ross.abutalabs.com/products/trulens
Homepage: https://www.trulens.org/
Language: Python
License: MIT
License Family: permissive
Topics: machine-learning, neural-networks, explainable-ml, llmops, ai-monitoring, ai-observability, evals, llm-evaluation, llms, ai-agents, llm-eval, agentops, agent-evaluation
Last push: 2026-08-25T21:50:33+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 86, longevity 100
- inputs: {"age_days": 2130, "days_push": 8, "days_rel": 13, "gap_med": 6.0, "n_releases_24m": 69}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3525, forks 331 (observed 2026-08-28T04:08:08.058446+00:00)

## What it is
TruLens is a Python library for evaluating and tracking LLM-based applications and AI agents. It provides OpenTelemetry-native tracing of every step (latency, tokens, cost) plus LLM-as-a-judge feedback metrics like the RAG triad, with version comparison and leaderboards.

## Use cases
- evaluate rag pipeline quality with groundedness metrics
- trace and debug why my ai agent gave a bad answer
- compare llm app versions on cost latency and quality
- run llm as a judge evals on agent traces
- monitor token usage and cost per step of llm calls
- catch agent errors automatically with eval metrics
- benchmark rag evaluation metrics against human annotations

## When to choose
- you need systematic evaluation and observability for LLM apps or agents in Python
- you want OpenTelemetry-native traces portable to any OTLP backend
- you need benchmarked RAG triad or agent error-detection metrics out of the box

## When to avoid
- you need a fully managed hosted LLM observability platform rather than an open-source library
- your stack is not Python-based
- you only need simple logging without evaluation metrics

## Facets
- artifact type: library
- maturity: active
- function: monitoring, tracing, testing, benchmarking, machine-learning, llm-inference, rag, agent-framework, analytics
- domain: large-language-models, machine-learning, developer-tools, monitoring, data-science
- platform: python, cross-platform
- tags: llm-evaluation, llmops, agentops, ai-observability, opentelemetry, llm-as-a-judge, rag-triad, evals, tracing, ai-agents, retrieval-augmented-generation

## Member repositories
- truera/trulens (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:08.058446+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:35:10.600008+00:00, confidence not recorded.
  - readme: https://github.com/truera/trulens (fetched 2026-08-28T04:08:08.058446+00:00, sha 6eb831982069)
  - homepage: https://www.trulens.org/ (fetched 2026-08-29T09:29:17.664594+00:00, sha 1cd84f0c3bf7)
  - site_page: https://www.trulens.org/getting_started/quickstarts/quickstart (fetched 2026-08-29T09:29:17.675890+00:00, sha 9c7c68e91561)
  - site_page: https://www.trulens.org/getting_started/install (fetched 2026-08-29T09:29:17.677376+00:00, sha 8f1185a15dc5)
  - registry_pypi: https://pypi.org/pypi/trulens/json (fetched 2026-08-29T09:29:17.678828+00:00, sha d046ce8c5cc9)
  - site_page: https://www.trulens.org/getting_started (fetched 2026-08-29T09:29:17.674132+00:00, sha 3e32c1b5e80c)
- Data as of 2026-08-30T08:39:29.467469+00:00.