Does the feedback system share my private source code?

No. The system is designed to block code contents, file paths, and credentials by design. Only aggregated, non-identifiable usage metrics are shared if you explicitly opt-in.

How do I stop Claude from learning my command patterns?

You can use the `/feedback pause` command to temporarily suspend learning, or `/feedback settings` to disable specific learning features permanently.

Will resetting my feedback data delete my project files?

No, the `/feedback reset` command only clears learned AI patterns, agent metrics, and permission rules stored in the .claude directory; it does not affect your project code.

Can I see the analytics data before it is shared?

Yes, the `/feedback export-analytics` command generates a JSON file of all metrics intended for sharing, allowing you to review every data point locally.

Feedback & Learning Manager

Name: Feedback & Learning Manager
Author: yonatangross

byyonatangross

•

分析と監視

Optimizes AI performance by managing usage analytics, learned patterns, and privacy settings within the OrchestKit ecosystem.

The Feedback skill is the central control hub for OrchestKit's self-optimizing capabilities. it allows Claude to learn from your specific coding patterns, preferred styles, and command approvals to reduce friction in future sessions. Beyond efficiency, it provides a transparent management interface for privacy, allowing you to view learned data, export analytics, and control GDPR-compliant consent settings. Whether you're tracking agent success rates or resetting learned behaviors, this skill ensures your AI environment evolves according to your needs while maintaining strict data security.

主な機能

01GDPR-compliant privacy controls with explicit opt-in/opt-out

02Detailed agent performance tracking and success metrics

03Automated learning of command patterns and preferred code styles

04Local data management including reset, export, and status viewing

05Security-hardened auto-approval rules for safe command execution

0669 GitHub stars

ユースケース

01Auditing the success rates of different AI agents to optimize development workflows

02Reviewing and exporting anonymous usage data for organizational privacy compliance

03Training Claude to recognize and auto-approve project-specific build and test commands

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit feedback

For use in Claude.ai and ChatGPT

Download Skill