Make0 Brain OSS Launches Shared Local Agent Memory

// 2h agoOPENSOURCE RELEASE

Make0 Brain OSS Launches Shared Local Agent Memory

ANNOUNCEMENT PRODUCT GITHUB PRODUCT HUNT

Make0 Brain OSS is a free, MIT-licensed local MCP memory server that lets Claude Code, Cursor, Codex, Windsurf, and other agents share context through one offline SQLite store. It supports versioned memory changes and local operation without an API key for core functionality. [Product Hunt](https://www.producthunt.com/products/make0-brain) [GitHub](https://github.com/make0-official/brain-oss)

// ANALYSIS

Make0’s strongest idea is treating memory as user-owned infrastructure instead of an isolated feature inside each agent, though local ownership also makes backups and security the user’s responsibility.

–Shared MCP access eliminates repeated project explanations across coding agents.
–Versioned, attributed edits and archive-on-delete behavior make multi-agent writes more recoverable.
–Offline MiniLM retrieval keeps sensitive context on-device and reportedly reaches 92.9% top-10 recall on SWE-ContextBench Lite. [GitHub](https://github.com/make0-official/brain-oss)
–The local SQLite database is not encrypted by the application and has no automatic backups, so users must manage disk protection and exports.
–The project targets a growing agent-memory layer, competing conceptually with Mem0, LangMem, and hosted memory platforms.

// TAGS

make0-brainagent-memorymcpopen-sourcelocal-firstcontext-engineeringdevtool

DISCOVERED

2h ago

2026-11-06

PUBLISHED

7h ago

2026-11-06

RELEVANCE

9/ 10

AUTHOR

Saurabh Singh

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

RESEARCH36m ago

Coca-Cola Compresses Tool Outputs Into Latents

Tsinghua researchers’ Coca-Cola replaces verbose tool observations with continuous latent tokens injected directly into an agent’s embedding space. Its action-preserving training targets next decisions rather than text reconstruction, reducing token consumption by 70.01%-89.24% and runtime by 24.45%-55.97% across web-search QA benchmarks.

BENCHMARK1h ago

Liber-0 Hits 36% Zero-Shot Robot Success

LiberAI’s World Action Model achieved 36% success across 600 trials on 60 GPT-6-generated manipulation tasks without task-specific robot demonstrations. It reached 63.5% on pick-and-place, 34% on short manipulation, and 10.5% on challenging multi-step tasks. ([LiberAI](https://www.liberai.com/blog/liber0/))

RESEARCH1h ago

Route to Remember Makes Weight Memory Practical

Amazon researchers show that on-policy self-distillation can make per-user weight memory outperform text retrieval for factual recall, reaching 59.9% accuracy versus 47.8% for retrieval on HaluMem. A verify-then-regenerate router combining both approaches raises accuracy to 66.1%.