search_memory - MemoryBase

The most important MCP tool. Your coding assistant uses this to find conversations and project context that are relevant to what you’re working on. It uses semantic search (meaning-based, not just keywords) — the same search that powers the browser extension.

Parameters

Parameter	Type	Required	Description
`q`	string	Yes	The search query (max 2,000 characters)
`topLayers`	number	No	Max layers to return (1-10)
`projectIds`	string[]	No	Only search within these project IDs (max 100)
`pinnedProjectIds`	string[]	No	Always include these projects in scoring (max 50)
`render`	boolean	No	If `true`, returns a ready-to-use markdown context pack

What It Returns

Ranked results containing:

Conversation summaries and context snippets
Project layers that match the query
Relevant entities (people, tools, concepts) mentioned across your history

When render: true, returns a single markdown document your assistant can use directly for context.

When Your Assistant Uses This

You ask about something you’ve discussed before
You reference a past decision or conversation
Your assistant needs background context for a task
You say things like “remember when we discussed…” or “use the approach from last week”