load-memory
Reload memory context from files
When & Why to Use This Skill
The 'load-memory' skill enables Claude to dynamically synchronize its internal context with external documentation files. It allows the agent to ingest updated project overviews, architectural decisions, and structured facts from local files like project.md and facts.json, ensuring the AI's responses remain aligned with the latest project state without requiring a session restart.
Use Cases
- Post-Manual Documentation Updates: Quickly refresh Claude's understanding after manually editing project.md, architecture.md, or coding conventions to ensure the agent follows the newest guidelines.
- Context Refresh for Long Sessions: Use this skill to clear 'stale' context and reload the most relevant facts and rolling memory when the agent begins to provide outdated or inconsistent information.
- Knowledge Verification: Execute the command to audit and verify exactly what information is currently loaded into the agent's memory, ensuring all critical project constraints are active.
| name | load-memory |
|---|---|
| description | Reload memory context from files |
Load Memory
Reload memory context into current session.
Usage
/memory-keeper:load-memory
Actions
Run the load-memory script:
node "scripts/load-memory.js"
This will output the current memory state to context:
Hierarchical Memory (if exists):
project.md- Project overviewarchitecture.md- Architecture decisionsconventions.md- Coding conventions
Rolling Memory:
- Last 50 lines of
memory.md
- Last 50 lines of
Structured Facts:
- Recent decisions, patterns, issues from
facts.json
- Recent decisions, patterns, issues from
When to Use
- After manually editing memory files
- To refresh context if it seems stale
- To verify what memory is currently loaded
Notes
- Memory is automatically loaded on session start
- This command reloads without restarting session