What are the confidence scores used for?

Confidence scores help determine the 'autonomyLevel' of the AI, indicating how much supervision Claude needs for specific types of tasks.

What is the difference between seed and personalized playbooks?

Seed playbooks are default instructions, while personalized playbooks are learned behaviors based on your specific coding style and feedback.

Where is the ALM status data stored?

All data is stored locally in your ~/.claude/alm/ directory, ensuring your learning history and metrics remain private.

How do I know if Claude needs a reflection session?

The dashboard flags 'stale' playbooks and displays a 'Ready for reflection' section when correction rates rise or data becomes outdated.

What does the ALM Status skill track?

It tracks session counts, confidence scores, autonomy levels per task, classifier training status, and the health of your personalized playbooks.

ALM Status Dashboard

Name: ALM Status Dashboard
Author: sfw

bysfw

0•

分析と監視

Visualizes the learning progress, confidence metrics, and health of the Claude Code Adaptive Learning Memory system.

The ALM Status skill provides a comprehensive diagnostic dashboard for the Adaptive Learning Memory (ALM) system, allowing users to track how Claude evolves over time. It aggregates data from local session logs to display autonomy levels, confidence scores for specific task types, and the training status of the underlying classifier. By identifying stale playbooks with rising correction rates, it helps users determine exactly when to trigger reflection cycles to optimize Claude's task performance and reliability.