← All repos

helicone

🧊 Open source LLM observability platform. One line of code to monitor, evaluate, and experiment. YC W23 🍓

agent-monitoringanalyticsevaluationgptlangchainlarge-language-modelsllama-indexllmllm-costllm-evaluationllm-observabilityllmopsmonitoringopenaiopen-sourceplaygroundprompt-engineeringprompt-managementycombinator
Browse cluster: Claude AI Agent Frameworks & MCP Tools
5,479commits
105contributors
9languages

Tech stack & purpose

Helicone is an open-source LLM observability platform built as a monorepo using Node.js (TypeScript) and Yarn workspaces. The project provides monitoring, analytics, and management tools for large language model applications, with a tech stack that includes Next.js for the web frontend, Express.js for the backend API (Jawn), Cloudflare Workers for request proxying, and Rust for an AI gateway component. The platform uses PostgreSQL (via Supabase) for application data, ClickHouse for analytics, and integrates with LLM providers like OpenAI and Llama Index. The project is backed by Y Combinator (W23 batch) and includes tooling for prompt engineering, cost tracking, evaluation, and LLMOps workflows.

Community & reference links

Languages

TypeScript
91.2%
MDX
7.1%
Python
0.7%
PLpgSQL
0.3%
Shell
0.2%
JavaScript
0.2%
HCL
0.2%
CSS
0.1%
Dockerfile
0.0%

Contributors (top 30 of 105)