remember

Remember — Save to Super Memory

Save the provided information to the local semantic memory store.

Arguments

$ARGUMENTS

Instructions

1. Parse the arguments to extract:

   • content: The main text to remember (required — everything before any flags)
   • agent: Agent/user ID (flag --agent <id>, default: "default")
   • type: Memory type (flag --type <type>, default: "fact")
     • fact — factual information
     • preference — user preferences and habits
     • experience — things that happened
     • task — tasks and todos
     • context — project/session context
   • importance: Float 0.0–1.0 (flag --importance <n>, default: 0.7)
   • tags: Optional comma-separated tags (flag --tags <t1,t2>)

2. If no flags are provided, treat the entire argument as content and use smart defaults:

   • Detect memory type from content: preferences → preference, tasks/todos → task, etc.
   • Set importance based on urgency words ("important", "critical", "always") → 0.9+

3. Call mcp__super-memory__memory_store with the parsed values.

4. Confirm success with the returned memory ID and a short summary.

Examples

/remember User prefers Python over JavaScript
/remember --agent project1 --type preference --importance 0.9 Always use async/await, never callbacks
/remember --type task --importance 0.8 Review PR #42 before Friday
/remember --type context --tags docker,infra The production DB runs on port 5433 not 5432

Output format

After saving, respond with:

Saved to memory: "<first 60 chars of content>"
ID: <uuid>
Type: <type> | Importance: <score>

If the memory server is unreachable, say so clearly and suggest running: docker compose up qdrant -d and uvicorn app.main:app in D:\work\mnemoforge