EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Hugging Face Blog·Sep 3·5 min read
Give Your Coding Agents a Memory You Own
TL;DR
Coding agents generate valuable session logs that capture their decision-making processes, but these logs require better indexing and retrieval methods to be truly useful.
✦ Why It Matters
Start implementing better indexing and retrieval systems for coding agent logs to enhance their utility today.
Key Takeaways
- Install funes with a single command to enhance agent memory.
- Use funes to enable agents to recall past decisions automatically.
- Sync your memory across multiple machines for seamless workflow.
- Leverage funes for team collaboration by sharing decision histories.
How It Works
funes creates a memory layer by indexing session logs from coding agents, allowing them to recall specific past decisions and their context. It uses a deterministic pipeline to parse and embed these logs into a local Lance dataset, enabling efficient retrieval through a combination of vector and BM25 search methods.