mind-palace

Navigate Mind Palace to query entity knowledge about codebase subsystems. Use when understanding architecture, exploring how systems work, or checking for hazards/gotchas before modifying code.

⭐2stars🔀0forks📁View on GitHub🕐Updated Jan 1, 2026

When & Why to Use This Skill

The Mind Palace skill is a sophisticated knowledge management framework designed to provide AI agents with deep, structured insights into complex codebase architectures. It organizes system knowledge into 'Keeper' entities that store not just the 'how' of a subsystem, but also the 'why' behind design decisions and critical 'watch-out' warnings. By utilizing a deterministic slug system and hierarchical YAML structures, it enables agents to maintain persistent, high-fidelity context across different modules and development sessions, ensuring safer and more informed code modifications.

Use Cases

  • Architectural Onboarding: Rapidly understand the hierarchy and responsibilities of various subsystems (e.g., MemoryKeeper, SchedulingKeeper) when entering a new or complex codebase.
  • Pre-Modification Risk Assessment: Query specific entities before refactoring to identify hidden dependencies, 'gotchas,' and design constraints that aren't immediately obvious from the source code.
  • Design Rationale Retrieval: Access the underlying philosophy and 'why' of a system's design to ensure new features or fixes remain consistent with the original architectural intent.
  • Cross-Agent Context Synchronization: Use the deterministic slug system (region/building/room) to allow multiple specialized agents to reference the exact same codebase components without ambiguity.
namemind-palace
descriptionNavigate Mind Palace to query entity knowledge about codebase subsystems. Use when understanding architecture, exploring how systems work, or checking for hazards/gotchas before modifying code.
allowed-toolsRead, Glob, Grep

Mind Palace Navigation

Query keeper entities for codebase knowledge. Each entity guards a subsystem and knows HOW it works, WHY it's designed that way, and WATCH_OUT warnings.

Quick Reference

List entities: Read .mind-palace/palace.yaml for entity names

Query an entity: Read .mind-palace/entities/<entityname>.yaml (lowercase)

Available Keepers (22 total):

  • MemoryKeeper - ChromaDB, summaries, retrieval
  • SchedulingKeeper - Day phases, decision engine
  • AgentKeeper - Claude SDK, Temple-Codex, tool execution
  • SelfModelKeeper - Identity, observations, growth edges
  • GoalsKeeper - Unified goals, hierarchies, approval workflow
  • ConversationKeeper - Message persistence, threading
  • And 16 more...

Entity Format

Each entity YAML contains:

name: MemoryKeeper
slug: memorykeeper          # Deterministic ID for cross-agent refs
location: memory
role: "Guardian of hierarchical vector memory..."
topics:
  - name: semantic search
    how: "Vector-based memory using ChromaDB..."
    why: "Finding relevant context requires semantic understanding..."
    watch_out: "Attractor basins use specific marker format..."

Slug System

All palace elements have deterministic slugs for cross-agent communication:

  • Entities: memorykeeper, schedulingkeeper
  • Rooms: memory-add-message (file-function pattern)
  • Buildings: memory (file stem)
  • Regions: backend (directory)

Path format: {region}/{building}/{room} → backend/memory/memory-add-message

Slugs survive regeneration - same codebase produces same slugs.

Sub-Palaces

Each major directory has its own sub-palace:

  • backend/.mind-palace/ - Python backend
  • admin-frontend/.mind-palace/ - React admin UI
  • tui-frontend/.mind-palace/ - Textual TUI
  • mobile-frontend/.mind-palace/ - React Native mobile

Root .mind-palace/ contains shared entities (Keepers). Sub-palaces contain regions/buildings/rooms for their scope.

Usage Examples

Before modifying the scheduler:

  1. Read .mind-palace/entities/schedulingkeeper.yaml
  2. Check topics for "day phases", "phase queues", "decision engine"
  3. Note the WATCH_OUT warnings

Understanding memory system:

  1. Read .mind-palace/entities/memorykeeper.yaml
  2. Topics cover semantic search, hierarchical retrieval, journals, context sources

Finding the right entity:

  1. Grep .mind-palace/entities/*.yaml for keyword
  2. Read matched entity for full context

Files

  • .mind-palace/palace.yaml - Palace index with entity list
  • .mind-palace/entities/*.yaml - Individual entity definitions
  • backend/mind_palace/ - Palace implementation code