Overview
Lunary is a toolkit for developing and operating LLM chatbots. It provides conversation and feedback tracking, cost and latency analytics, debugging logs, prompt directory versioning, and evaluation datasets.
Key features
- Conversation & feedback tracking with prompt directory and versioning.
- Cost, token, and latency analytics to optimize invocation strategies.
- Debugging and observability (logs, traces) and SDKs for JS/Python with LangChain integrations.
Use cases
- Debugging, evaluation, and A/B testing for chatbot development.
- Team prompt management, audit, and quality regression testing.
- Self-hosted or hosted observability and data collection for LLM services.
Technical notes
- Offers both hosted and self-hosted deployment options, uses PostgreSQL for storage, and integrates with common LLM frameworks and SDKs.