Hugging Face details Funes: turning agent session traces into durable, searchable memory
ariG23498 · x · 2026-09-07
Hugging Face publishes a deep-dive on Funes, a durable, local memory layer for coding agents, supporting Claude Code, Codex, pi, and Hermes.
- Problem: across machines and agents, every session starts from zero; context like "why did we move off the streaming parser?" vanishes when sessions end.
- Diagnosis: agents already leave rich traces (not just what changed but why), but traces are only potential memory — you can't grep ten thousand turns; retrieval needs indexing, ranking, and exact provenance.
- Solution: Funes is a single binary with no ML runtime dependency; embedding and reranking run locally. funes add claude builds the first index, gives the agent recall/get tools, and auto-indexes each completed turn.
- Portability: memory stays local and private by default, and can sync to a Hugging Face dataset you own.
The article operationalizes the earlier "Software Forgets: Agent Traces Are the Memory" thesis, turning archives into memory agents can actually use.
More from coding & agent
- Resin lets an agent build standalone LLM inference engines; vLLM and SGLang optimize DeepSeek kernels differently — AccBalanced · 2026-10-27
- How I let my cloud Grok Bot code on my Mac mini via Tailscale, no public ports — talkaboutdesign · 2026-10-27
- Context layout is a cost decision: one memory edit can bust your entire agent prompt cache — dscape · 2026-10-27
- Alchemy adds effect/Workflow support across Durable Objects, Workflows and Lambda Durable Functions — samgoodwin89 · 2026-10-27
- Alchemy's Lambda Web Functions run plain Node web servers on Lambda, no OpenNext needed — samgoodwin89 · 2026-10-27
- Alchemy Domain brings domain-driven design and story testing to Effect backends — samgoodwin89 · 2026-10-27