Help us improve
Share bugs, ideas, or general feedback.
From Memsy
Recalls past context, decisions, and discussions from Memsy memory. Activates on retrieval-intent queries like "what did we decide" or "search past conversations."
npx claudepluginhub memsy-io/memsy --plugin memsyHow this skill is triggered — by the user, by Claude, or both
Slash command
/memsy:memsy-recallThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
The user is asking you to recall something from Memsy memory. Run this workflow:
Searches and surfaces relevant memories from past sessions to inform current work with decisions, patterns, and learnings.
Searches Mem9 shared memories from past sessions via Bash API calls to retrieve relevant historical context, decisions, project knowledge, or team expertise.
Implements 3-layer memory search workflow to recall past work, decisions, errors, and project history token-efficiently via layered functions.
Share bugs, ideas, or general feedback.
The user is asking you to recall something from Memsy memory. Run this workflow:
From the user's message, strip framing words and keep the substantive nouns:
| User said | Extract |
|---|---|
| "What did we decide about billing storage?" | billing storage |
| "Do we have anything on the auth migration?" | auth migration |
| "Remember when we picked Postgres over Mongo?" | Postgres Mongo decision |
| "Context on the rate limiter rewrite" | rate limiter rewrite |
If the message is ambiguous (e.g. "remember that?" with no antecedent), ask a clarifying question before calling the tool. Don't search for an empty / one-word query unless the user actually meant a single topic.
query: the extracted topiclimit: 8Format as bullet points, grouped by theme when 3+ results share a thread (e.g. all about the same migration). Each bullet uses the fields memsy_search returns (id, score, content, metadata):
content (truncate to 200 chars)Top 3–5 most relevant first. If memories are clearly unrelated to the query (low scores, off-topic), surface only what's relevant — don't pad. Do not invent fields like observed_at or actor_id — search results don't carry them at the top level; call memsy_get_memory for a specific result if the user needs more detail.
Say clearly: "No memories matched <query>." Then offer:
/memsy-doctor — the memory might be in a different org./memsy <query> directly with different wording.Hand off to the memsy-setup skill — do not retry blindly, do not fabricate substitute answers. Be explicit that Memsy isn't reachable right now and point at /memsy-doctor.