Skip to content
ɳSelf Docs

Memory-Rooms

Feature: Memory Rooms

Section titled “Feature: Memory Rooms”

Memory Rooms let you organize ɳClaw’s knowledge into named collections. Think of them as folders for your AI’s memory. You can drag and drop memories between rooms, and each room tracks its own health metrics.

Route: /settings/memory in claw-web API: Memory rooms CRUD under /claw/memory/rooms

How It Works

Section titled “How It Works”

Every memory in ɳClaw belongs to a room. The default room holds unorganized memories. You create named rooms for topics you care about (work, health, projects, people) and assign memories to them.

The room view shows:

  • Memory count per room
  • Health score (based on freshness, connections, usage)
  • Drag-and-drop reordering
  • Bulk assignment of memories

API

Section titled “API”

List Rooms

Section titled “List Rooms”
GET /claw/memory/rooms
Authorization: Bearer {token}

Create Room

Section titled “Create Room”
POST /claw/memory/rooms
Content-Type: application/json
Authorization: Bearer {token}


{
  "name": "Work Projects",
  "description": "Memories about active work projects"
}

Assign Memories

Section titled “Assign Memories”
POST /claw/memory/rooms/{id}/assign
Content-Type: application/json


{
  "memory_ids": ["mem_abc", "mem_def"]
}

Update Room

Section titled “Update Room”
PUT /claw/memory/rooms/{id}

Delete Room

Section titled “Delete Room”
DELETE /claw/memory/rooms/{id}

Deleting a room moves its memories back to the default room.

Knowledge Health

Section titled “Knowledge Health”

The /knowledge route shows a brain health report across all rooms. It checks for:

  • Stale memories (not accessed in 30+ days)
  • Orphan memories (no connections to other knowledge)
  • Duplicate or near-duplicate entries
  • Low-confidence facts that need verification

The health check runs via GET /claw/knowledge/health.

Ingestion Pipeline

Section titled “Ingestion Pipeline”

The auto-ingestion system imports knowledge from external sources:

  • URLs (web pages, articles)
  • Files (PDF, Markdown, text)
  • Clipboard content

Monitor progress at /settings/knowledge/ingestion. The ingestion queue is backed by np_claw.ingestion_queue.

Export

Section titled “Export”

Export your knowledge graph in two formats:

  • Obsidian — Markdown files with wiki-style links
  • JSON — Full structured export

GET /claw/knowledge/export?format=obsidian

Section titled “Related”