Browse

browser-use

Python library that makes web browsers accessible to AI agents; built on Playwright and LangChain. Supports multi-tab, vision + accessibility-tree hybrid mode, custom actions, and a self-correcting agent loop.

daytona

Secure and elastic sandboxes for running AI-generated code. The public repository stopped updating in June 2026, when development moved to a private codebase.

ggml-org-llama-cpp

LLM inference in C/C++

Contributed by Sentinel

vllm-project-vllm

A high-throughput and memory-efficient inference and serving engine for LLMs

Contributed by Sentinel

e2b-sandbox

Use as the default runtime when an agent must execute untrusted code or commands. Firecracker microVMs with ~150ms cold start give each run an isolated, disposable computer.

diegosouzapw-omniroute

Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors

Contributed by Sentinel

sgl-project-sglang

SGLang is a high-performance serving framework for large language models and multimodal models.

Contributed by Sentinel

sentry-llm-monitoring

Sentry's error and performance monitoring extended to LLM applications. It captures exceptions, latency, and AI token usage with OpenTelemetry integration.

stripe-agent-toolkit

Use as the payments rail when an agent should earn or spend money in code (create customers, prices, payment links, and usage-based billing): the infrastructure behind an agent that funds its own compute.

mlflow-tracing

MLflow's LLM tracing module instruments model calls, agent steps, and tool invocations, storing them alongside experiment runs for reproducibility.

langfuse

Open-source LLM engineering platform with traces, evals, prompt management, and datasets for debugging and improving LLM applications.

promptfoo

CLI and library for testing, comparing, and red-teaming LLM prompts and agents with assertions and CI integration.

claude-hud

A status line for Claude Code that shows context usage, tools, agents, to-dos and more. Highly configurable, and maintained when it was listed.

skyvern

Open-source agent platform that automates browser-based workflows using LLMs and computer vision. It identifies interactive elements via screenshots, handles CAPTCHAs, and supports complex multi-step form flows.

openai-evals

OpenAI's official framework for evaluating LLMs and LLM-powered systems, with a registry of community eval sets.

lm-evaluation-harness

EleutherAI's unified framework for evaluating language models on hundreds of academic benchmarks.

browserbase-bb

Use when an agent must operate the live web (navigate, act, and extract on real pages) via a cloud browser driven by act/extract/observe primitives, with a local-Chromium escape hatch using the same code.

comet-opik

Opik by Comet is an open-source LLM evaluation and tracing platform: log traces, run automated evals, create datasets, and track prompt improvements over time.

browser-use-webui

Gradio web UI on top of the browser-use framework. It lets users run AI browser agents interactively, configure LLM providers, watch live recordings, and replay task sessions without writing Python.

opentelemetry-genai

OpenTelemetry semantic conventions and instrumentation for GenAI/LLM spans, traces, and metrics via the GenAI semconv working group.

cua-computer-use-agent

trycua/cua open-source computer-use agent framework: Apple Silicon-native, runs lightweight macOS/Linux VMs with sub-second cold starts; provides a unified Python interface for screen capture, click, and type actions.

deepeval

Open-source LLM evaluation framework with 14+ metrics (hallucination, faithfulness, answer relevancy) and CI support.

ragas

Reference-free evaluation of retrieval-augmented generation pipelines; measures faithfulness, answer relevance, and context precision.

mcp-tunnels-cloudflared

Use to let a hosted agent reach a private-data MCP server behind your firewall (an outbound tunnel plus a proxy, with per-server OAuth) so internal tools are usable without exposing them to the public internet.