Hugging Face Open-Sources Funes: A Persistent, Cross-Machine Memory Layer for Coding Agents
On September 7, Hugging Face engineer @ariG23498 released and detailed Funes, an open-source project that gives coding agents a local, persistent memory layer, directly addressing the pain point of "forgetting everything when you switch machines or sessions." The project drew community attention and shares on the day of its release.
Confirmed
- Funes was developed and open-sourced by Hugging Face's engineering team; @ariG23498 is the original poster, and users like @MaziyarPanahi helped spread the word
- Core positioning: a one-stop agent memory storage solution that saves an agent's decision trajectories and retrieves them on demand
- Supports fully compaction-free operation, persistent memory across machines, and switching agents mid-session
- Compatible with mainstream coding agent tools like Claude Code, Codex, pi, and Hermes, and can build memory from existing session logs
- @MaziyarPanahi added that git cannot solve this problem: ignored files, system-level skills, and the like live outside the repository, which is why a dedicated memory-sync solution is needed
Why it matters
- Coding agents today generally start from scratch in every new session, losing all context across machines and in multi-agent collaboration—one of the most widely acknowledged pain points
- Funes turns session trajectories into retrievable, persistent assets, with local storage and no vendor lock-in, offering direct practical value for multi-machine workflows and long-term project maintenance
2026-09-07 ~ 2026-09-07 · 6 related posts
Primary sources
- [source] Hugging Face open-sources Funes, a durable memory layer for coding agents — ariG23498 · 2026-09-07
- [source] Hugging Face details Funes: turning agent session traces into durable, searchable memory — ariG23498 · 2026-09-07
- Hugging Face unveils funes, a persistent memory layer for coding agents — ariG23498 · 2026-09-07
- [source] Git can't sync agent memory: Hugging Face's funes tackles cross-machine agent state — MaziyarPanahi · 2026-09-07
2 near-duplicate retellings: ariG23498 · MaziyarPanahi