Install with AI
Your coding agent can integrate RememberOS for you — paste one prompt, or point it at MCP.
One-prompt setup (Claude Code, Cursor, etc.)#
Paste this into your agent in the project you want memory-enabled:
Use the RememberOS memory API at https://rememberos.ai. Read the bearer key from the
LONGMEM_API_KEY env var. Before answering, search relevant memory with
POST /v1/memory/search {"query": <user intent>}. After learning a durable fact
about the user or project, store it with
POST /v1/memory/collections/project/memories {"text": <fact>}. Prefer the
longmem Python or JS SDK if available.The agent reads llms.txt — the machine-readable API summary at the site root — and wires the calls itself.
MCP (Model Context Protocol)#
RememberOS speaks MCP over HTTP: point any MCP-capable agent at
POST https://rememberos.ai/v1/mcp with your mv_… key as a bearer
token. Available tools:
| Tool | Does |
|---|---|
search_memory | hybrid search across your memories |
store_memory | store a new memory |
get_memory | fetch one memory by id |
list_collections | list your collections |
related_memory | find memories related to one — discover patterns |
what_changed | where newer facts superseded older ones |
collection_health | stale / expired / supersessions + needs_attention |
summarize_collection | board-report briefing — summary + key themes |
find_duplicates | surface near-duplicate clusters (read-only) |
archive_stale | soft-archive stale memories (reversible, dry-run default) |
What the agent should know#
- Collections are namespaces — use one per project or per end-user.
- Container tags sub-scope inside a collection (e.g. one per end-user of your app).
remember(fact extraction) costs LLM calls;addis free of LLM cost.- Search before answering; store after learning something durable.