Cerebrun
Unified memory layer for AI assistants. Built on MCP and pgvector for persistent, searchable context across conversations.
Cerebrun: Launched. Get it at cereb.run.
User Context MCP is a persistent identity and memory layer for people who work across several AI agents such as Claude, ChatGPT and Windsurf. It stores context in four layers: public preferences like language and tone, active goals and pinned memories, personal details for trusted agents, and a vault for passwords and API keys. Only preferences are injected by default, and agents use a search_context tool to pull deeper context on demand. Vault access requires per-session approval and a temporary token, every read is logged, and an LLM gateway lets agents search, fork and continue conversations across providers.
Company details
Lifeline
▣ GraduatedEngagement
Get the week's earliest tools in your inbox.
One email every Sunday: new launches, staff picks and early-access deals.