Store and recall information in hAIveMind using natural-language requests. Use when the user says things like remember this, save this, don't forget, recall, what do you remember, or asks to retrieve prior decisions, context, or notes.
Lets you store and recall anything by just saying it. No commands, no syntax. Just talk to Jarvis like he has a memory — because now he does.
"Jarvis, remember that the staging server password is rotating on Friday."
"Note that Andrew prefers email over Slack for status updates."
"Don't forget — client prefers morning calls before 10am."
"What do you remember about the project-beta project?"
"Recall what I said about the deployment process."
Uses the hAIveMind MCP server — a vector database that stores memories as searchable embeddings. Memories persist across sessions, restarts, and compactions.
Storing:
mcporter call haivemind.store_memory \
content="[what was said]" \
category="[appropriate category]"
Retrieving:
mcporter call haivemind.search_memories \
query="[what they're asking about]" \
limit=10
Jarvis automatically categorizes based on content:
global — general notes, preferences, decisionsrules — standing instructions ("always CC John on client emails")operations — deployments, schedules, processesinfrastructure — servers, services, credentials (handle carefully)development — code decisions, PR notes, bug contextAlways stored with timestamp and context:
[ISO timestamp] [topic]: [content]
Memories are retrieved by semantic similarity — you don't need to remember exact phrasing to recall them.
Requires hAIveMind MCP server configured in OpenClaw.
See SETUP.md for installation instructions.