Memory is where your AI writes down what it learns and reads it back before the next task. Every session starts smarter than the last, in every tool you use, for everyone on your team.
You re-describe the project, the stack and the standards to every tool, every time.
What your AI worked out yesterday is gone today, and you become its memory, copying context from chat to chat.
Claude's memory does not help in Cursor. Context is trapped per app and per account, never shared with your team.
Before a task your agent searches memory by meaning and pulls the relevant context. As it works, it writes new lessons back. VibeXP does not watch your sessions: your AI reads and writes deliberately.
vibexp_io_create_memory( text: "Postgres pool maxes out at 20 under load. pool_max 50 fixed the timeouts.", labels: ["postgres", "perf"] )
Semantic search finds the right memory when you do not remember the words, across memories, prompts, blueprints and artifacts. It is the same retrieval your agents call before they work.
vibexp_io_search(query: "why do requests time out?") # memory Postgres pool maxes out at 20... 0.93 # blueprint backend statement timeouts... 0.71
Two explicit steps over MCP. Before a task, your agent searches memory by meaning and pulls the relevant context. As it works, it writes new lessons back through the memory tools. The only automatic part is indexing: a memory is embedded the moment it is saved.
In the docs ↗Built-in memory lives inside one tool and one account. VibeXP memory is shared across every tool you use, and across everyone on your team, so context never gets stranded in one app.
In the docs ↗Anything worth remembering: a fact, a decision, a snippet, a preference, a lesson learned. A memory is free-form text plus metadata, scoped to a project, with no title required.
In the docs ↗Stars are how the next person tired of re-explaining their project to an AI finds this. It is the whole of the distribution: no ad budget, no sales team, no hosted plan to upsell.
Free, open source and self-hosted. One command to run it, one to connect your AI tools, and everything else in the docs.