Observability & Evals

LLM App is a Python library that helps you build real-time LLM-enabled data pipelines with few lines of code

OpenAI-compatible proxy that stacks the free tiers of 28 LLM providers (~4B free tokens/month) behind one /v1 endpoint — smart routing, automatic failover, encrypted keys

The most widely adopted self-hostable LLM observability platform: traces every agent step, manages prompt versions, and runs evals in one tool. Preferred over cloud-only alternatives when data residency or cost control is a constraint

Visualizer for neural network, deep learning, and machine learning models

YAML-driven LLM testing framework with LLM-as-judge, assertion DSL, and native CI integration. The most practical tool for adding agent output regression tests to a PR pipeline without writing a test harness from scratch

Comet's open-source AI observability and evaluation platform: deep tracing of LLM calls, conversation logging, and agent activity, plus built-in eval metrics, prompt versioning, guardrails, and the Opik Agent Optimizer. Worth including because it unifies observability, verification, and optimization in one self-hostable stack rather than stitching together separate tools

macOS menu bar app showing usage stats for Codex, Claude, Cursor, Gemini without login. On-device parsing

by ryoppippi - A zero-install CLI (npx ccusage) that analyzes Claude Code token usage and cost from local JSONL logs — daily, monthly, per-session, and 5-hour-block breakdowns, a live monitoring mode, and per-model cost estimates. Runs entirely locally, with JSON output for scripting

The most complete open-source LLM/agent eval framework: 20+ built-in metrics (hallucination, answer relevancy, RAGAs, tool correctness), pytest integration, and a CI-friendly runner. Removes the need to hand-roll eval infrastructure when you need structured, repeatable agent quality gates

RAG evaluation with knowledge-graph-based test set generation and 30+ metrics. ~8K+ ⭐

Python SDK for Agent AI Observability, Monitoring and Evaluation Framework. Includes features like agent, llm and tools tracing, debugging multi-agentic system, self-hosted dashboard and advanced analytics with timeline and execution graph view

A framework for few-shot evaluation of language models

Open source tool for automated code reviews. Qodo was formerly known as Codium (not to be confused with Codeium with an "E")

a high-performance, high-precision CPU, GPU, and memory profiler for Python

by sirmalloc - A highly customizable status line formatter for Claude Code CLI that displays model info, git branch, token usage, and other metrics in your terminal

Self-hostable trace UI and eval runtime for agent workflows. Lets harness engineers audit and replay every reasoning step and tool call offline, without sending data to a third-party cloud

Always know what to expect from your data

Free, local tool that tracks AI coding token usage and cost across 31 tools and agents by model, project, and task. Worth including because cross-tool cost visibility is the missing prerequisite for agent FinOps in multi-tool teams — most observability tools either require cloud upload or only cover a single provider

A developer first, lightweight, user-friendly experiment tracking and visualization tool for machine learning projects, streamlining collaboration and simplifying MLOps. W&B excels at tracking LLM-powered applications, featuring W&B Prompts for LLM execution flow visualization, input and output monitoring, and secure management of prompts and LLM chain configurations

Easily fine-tune (SFT/RL), evaluate, and deploy Qwen, Gemma, or any open agentic LLM/VLM!

(8.2k ⭐) - A real-time Claude Code usage monitor with predictions and warnings

Creating semantic cache to store responses from LLM queries

Evidently is ​​an open-source ML and LLM observability framework. Evaluate, test, and monitor any AI-powered system or data pipeline. From tabular data to Gen AI. 100+ metrics

Sweep is an AI tool that automates the transformation of GitHub issues into pull requests, streamlining code improvements and bug fixes, supported by a suite of features like codebase understanding, test running, and a developer-friendly interface for installation and usage github | website

Open-source, real-time cost observability platform for AI agents. Track tokens, costs, messages, and model usage with a local-first dashboard. Supports 28+ LLM models, OTLP ingestion, self-hosted

OpenCompass is an LLM evaluation platform, supporting a wide range of models (Llama3, Mistral, InternLM2,GPT-4,LLaMa2, Qwen,GLM, Claude, etc) over 100+ datasets

OpenTelemetry-based instrumentation for LLM calls and agent steps: adds trace spans to every inference and tool call without modifying business logic. The cleanest way to bring the existing OTEL ecosystem (Grafana, Datadog, Jaeger) to a harness

Observability and enforcement for AI agent harnesses. Capture every run and runtime reliability with policy enforcement. 40 built-in policies, a local dashboard, no account required with a generous free cloud plan

TensorFlow's Visualization Toolkit

Auto-Magical CI/CD to streamline your ML workflow. Experiment Manager, MLOps and Data-Management

by Ruller_Lulu - A desktop pet that reacts to your Claude Code sessions in real-time — thinking, typing, juggling, sleeping, and more. Yep. It's undeniably endearing. And at the end of the day, isn't that what Claude Code is all about?

非线智能 NoneLinear - ReLE评测：中文AI大模型能力评测（持续更新）：目前已囊括374个大模型，覆盖chatgpt、gpt-5.4、谷歌gemini-3.1-pro、Claude-4.6、文心ERNIE-X1.1、ERNIE-5.0、qwen3.6-max、qwen3.6-plus、百川、讯飞星火、商汤senseChat等商用模型， 以及step3.5-flash、kimi-k2.6、ernie4.5、MiniMax-M2.7、deepseek-v4、Qwen3.6、llama4、智谱GLM-5.1、MiMo-V2、LongCat、gemma4、mistral等开源大模型。不仅提供排行榜，也提供规模超200万的大模型缺陷库！方便广大社区研究分析、改进大模型。

(5.9k ⭐) - Local-first session search, analytics, insights, and token-use statistics for Claude Code, Codex, and more than 20 coding agents

an easy-to-use and performant open-source experiment tracker

Open-source LLM observability proxy (YC W23) with the largest open-source pricing database (300+ models). One-line proxy integration provides cost tracking, token monitoring, session tracing, and prompt versioning across providers. The AI Gateway component handles request routing and caching with zero-code changes. SOC 2 and GDPR compliant, self-hostable via Docker — the natural complement to exec

Benchmark for evaluating LLM systems on real-world GitHub issue resolution tasks

An open-source GPU cluster manager for running and managing LLMs

Open-source agent engineering platform (YC W24) with session replay, cost tracking, and failure detection across 10+ frameworks including CrewAI, LangGraph, and OpenAI Agents SDK. The step-by-step execution graph and cross-session metrics make it the most practical debugging layer for multi-agent systems in production

(3.8k ⭐) - Local API traffic viewer that attributes Claude Code and Codex token costs to prompts, tools, cache activity, and other request parts

CLI tool for tracking token usage from AI coding agents (OpenCode, Claude Code, OpenClaw, Codex, Gemini CLI, Cursor IDE, AmpCode, Factory Droid) with a global leaderboard and 2D/3D contribution graphs

An open platform for training, serving, and evaluating large language model for tool learning

A Python library that facilitates fast and easy data exploration by automating the visualization and data analysis process

Neural network 3D visualization framework, build interactive and intuitive model in browsers, support pre-trained deep learning models from TensorFlow, Keras, TensorFlow.js

LLM Ops platform with Analytics, Monitoring, Evaluations and an LLM Optimization Studio powered by DSPy

Open-source framework for prompt-learning research

The LLMOps platform to build robust LLM apps. Easily experiment and evaluate different prompts, models, and workflows to build robust apps

Open-source LLM observability and evaluation platform for AI agents. Traces agent runs, clusters production failures into issues, and auto-generates evals from real-world failures. Built on OpenTelemetry, self-hostable

Solve NLP Problems with LLM's & Easily generate different NLP Task prompts for popular generative models like GPT, PaLM, and more with Promptify

AI observability platform from the Pydantic team with a unique angle: all trace data is SQL-queryable (PostgreSQL-compatible), so coding agents can query production observability data directly via the Logfire MCP server. Full-stack OTEL tracing covers both the AI layer and backend — letting you determine whether a failure is in agent logic or infrastructure. The natural observability choice for Py

(4.4k ⭐) - macOS menu bar app for tracking Claude, Codex, and other AI coding subscriptions, with usage limits, reset times, credits, and spend

Sacred is a tool to help you configure, organize, log and reproduce experiments

LMQL is a Python-based programming language for large language models, allowing seamless integration of LLMs into code with advanced features like conditional logic, constraints, and multi-model support github | website

Tests for Continuous Validation of ML Models & Data. Deepchecks is a Python package for comprehensively validating your machine learning models and data with minimal effort

February 2026 open-source DevTools for Claude Code that reconstructs hidden session internals from local logs: per-turn token attribution across 7 context categories, full subagent execution trees with cost breakdowns, and syntax-highlighted diffs for every tool call. Essential because Claude Code's default UI deliberately collapses tool details and thinking steps — this tool restores the visibili

Multi-environment agent benchmark (OS, DB, web, code) with a structured eval pipeline. Worth studying for its environment isolation design and task definition format when building custom eval environments for your harness

Like htop but for AI coding agents. Monitor Claude Code and Codex CLI sessions, tokens, context window, rate limits in real-time

Evaluating and explaining LLM apps; tracks hallucinations, relevance, groundedness

by Haleclipse - A high-performance Claude Code statusline tool written in Rust with Git integration, usage tracking, interactive TUI configuration, and Claude Code enhancement utilities

(3,267 ⭐) - Desktop widget for tracking token usage, costs, limits, and session activity across Claude Code and other coding tools, with multi-device synchronization

Open-source all-in-one platform for engineering AI products. Traces, Evals, Datasets, Labels

Real-time architectural sensor that closes the feedback loop for coding agents: scans codebases, scores structural health, and surfaces degradation via MCP so agents can self-correct before entropy compounds. The gate --save / gate pair makes architectural regression detection CI-friendly — filling the verification gap between linters (style) and tests (behavior) for code produced by autonomous ha

(3,303 ⭐) - Local proxy and trace viewer for inspecting coding-agent API requests, prompts, tool calls, streaming responses, and token usage

Generates typed MCP servers, interactive API documentation, and SDKs from OpenAPI, AsyncAPI, GraphQL, gRPC, and OpenRPC sources

September 2026 agent-first linter for Tailwind design systems: you encode design-system contracts as programmable lint rules, and the errors carry your components' sizes, variants, and theme file paths so agents self-correct in one round instead of guessing. Validated across 150+ task runs (published evals: nearly all tasks reach zero violations in a single correction round, and lint feedback cost

[NeurIPS 2024] OSWorld - Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments

A python library for decision tree visualization and model interpretation

UK AI Security Institute's eval framework with native support for evaluating external agents (Claude Code, Codex CLI) as black-box targets, plus built-in bash/python/web browsing tools. Built for safety-grade rigor; the right foundation for harness-level eval infrastructure

Web framework to create vertical AI agents. FastAPI based, plugin system inspired to WordPress, admin panel, vector DB included

Open-source tools for testing and experimenting with prompts. The core idea is to enable developers to evaluate prompts using familiar interfaces like code and notebooks. In just a few lines of codes, you can test your prompts and parameters across different models (whether you are using OpenAI, Anthropic, or LLaMA models). You can even evaluate the retrieval accuracy of vector databases

Toolkit for creating, sharing, and using natural language prompts

Visual toolkit for building, testing, and comparing LLM prompt responses without code

(2,960 ⭐) - Desktop app that displays coding-assistant usage limits, reset times, and session status in a screen-edge notch, with macOS and Windows versions

OpenLIT is an OpenTelemetry-native GenAI and LLM Application Observability tool and provides OpenTelmetry Auto-instrumentation for monitoring LLMs, VectorDBs and Frameworks. It provides valuable insights into token & cost usage, user interaction, and performance related metrics

The open standard for data logging

NPM utility library for creating and maintaining prompts for LLMs (Microsoft)

Build and control your personal LLMs with fast and efficient fine-tuning

The de facto benchmark and execution harness for terminal agents: real sandboxed shell tasks verified by deterministic test scripts, now at 2.x on the Harbor framework. Cited throughout this list because it is the yardstick every harness-only improvement claims against — from LangChain's rank-30-to-top-5 harness changes to Self-Harness and Beagle's evolution loops — making it essential infrastruct

Seamlessly integrate LLMs as Python functions. Use type annotations to specify structured output. Mix LLM queries and function calling with regular Python code to create complex LLM-powered functionality

Evaluates coding-agent workflows across five dimensions and turns project and session evidence into prioritized, evidence-bounded findings with scoped repair actions. Runs inside Claude Code, Codex, Cursor, GitHub Copilot, and other hosts, making it a practical cross-tool diagnostic layer for improving the harness rather than just reviewing the final diff

Bionic is sovereign Agentic AI for the enterprise — Runs on-premise and can securely work with your sensitive data and systems.

by Nizar Selander - A hooks-driven system that monitors file operations in real-time and blocks changes that violate TDD principles

Cost-effective LLM development in any cloud (AWS, GCP, Azure, Lambda, etc)

(1.8k ⭐) - A local dashboard for tracking your Claude Code token usage, costs, and session history

Open-source, self-hostable platform unifying tracing, evals, simulations, guardrails, and gateway into a single feedback loop. Worth including because it demonstrates what a unified observability-and-improvement plane looks like rather than stitching together five separate vendor tools

Must-read papers and blogs on LLM based Long Context Modeling 🔥

Waku Waku! Waku Agent is a local-first AI agent harness you actually own, including loop, memory, eval, all in code built to stay legible as it grows.

ML models and internal tensors 3D visualizer

MLE-bench is a benchmark for measuring how well AI agents perform at machine learning engineering

Machine Learning automation and tracking

A model-agnostic visual debugging tool for machine learning

Web agent benchmark. Real websites

by tddworks - A macOS menu-bar app that surfaces remaining usage quota for Claude, Codex, Gemini, Copilot, and other AI coding providers at a glance, with burn-rate, dollar-balance, and reset-countdown indicators plus a lightweight live-session indicator. Swift 6 with a layered Domain/Infrastructure architecture

by disler - A real-time dashboard that captures Claude Code hook events across concurrent agents — tracing every tool call, task handoff, and lifecycle event through a Bun/SQLite/WebSocket/Vue stack, with session tracking and live filtering

Intuitive convenience tooling for lightning-fast, efficient development and ensuring quality in LLM-based applications

：创建、测试、度量和改进 Agent Skills

Benchmarks agent behavior in three-way user-tool-policy interactions — the failure mode SWE-bench doesn't cover. Useful for validating that a harness correctly enforces business rules across multi-turn, stateful conversations

Self-hosted, OpenAI-compatible LLM router. Bring your own provider keys and route across 100+ models from multiple providers with automatic failover and streaming; model="auto" selects a model by cost, latency, or quality. Includes a local analytics dashboard and does not send telemetry

Gym-style benchmark and environment toolkit for evaluating browser-using web agents

：Agent Skill 评测与演化工具

Replays Claude Code and Codex sessions on a 3D map of your codebase, turning raw JSONL logs into a spatial view of where the agent searched, read, and edited. The glow-based visualization makes exploration drift and context pressure immediately visible — a concrete debugging primitive for reviewing whether an agent's footprint matched the intended task scope