EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Hacker News·Jul 23·5 min read
Show HN: Claude-thermos – keeps your Claude session warm for you
TL;DR
Claude-thermos is a tool designed to keep your Claude AI session active, preventing prompt cache expiration during long interactions. By maintaining the session, it can reduce costs associated with re-encoding conversations.
✦ Why It Matters
Implement Claude-thermos today to cut down on unnecessary costs during long AI sessions.
Key Takeaways
- Use Claude-thermos to keep your cache warm during long sessions.
- Customize warming intervals to fit your usage patterns.
- Monitor your session logs for detailed cost savings.
- Ensure you have Python 3.11+ and the Claude CLI installed.
How It Works
Claude-thermos operates as a local reverse proxy that tracks session activity. It detects when the main agent is idle while a subagent is active and sends warm requests to refresh the cache.
These requests are lightweight, ensuring that the cache remains active without disrupting the main traffic. By doing so, it prevents the cache from expiring and avoids costly re-encodings.