Does this skill require manual approval for every change?

The skill offers multiple modes: users can choose to review every signal, apply all signals for a specific artifact once, or set a preference to remember for future sessions.

How does the skill handle conflicting feedback signals?

It resolves contradictions by prioritizing the most recent observations and the highest frequency reports, and escalates to the user if a mechanical resolution isn't possible.

How is the feedback evidence validated?

The skill performs sanity checks to verify target paths, signal quality (specificity), and duplicate detection before presenting any changes to the user.

What is the PAS framework mentioned in this skill?

PAS stands for Process-Agent-Skill, a modular architecture for building AI systems. This skill manages the 'improvement loop' by processing feedback signals within that framework.

What are Stability Anchors (STAs)?

Stability Anchors are protected behaviors or constraints within a PAS artifact. This skill checks every proposed change against STAs to ensure improvements don't break core functionality.

PAS Feedback Orchestrator

Name: PAS Feedback Orchestrator
Author: ZoranSpirkovski

byZoranSpirkovski

0•

Productivity & Workflow

Automates the review and implementation of accumulated feedback signals to iteratively improve PAS processes, agents, and skills.

The Applying Feedback skill is a core component of the Process-Agent-Skill (PAS) framework, designed to close the loop between agent execution and system optimization. It systematically scans backlogs for feedback signals—observations recorded by agents during self-evaluation—and applies pattern analysis to identify high-confidence improvements. By performing sanity checks against Stability Anchors (STAs) and prioritizing instruction consolidation over expansion, this skill ensures that your AI agents evolve safely and efficiently without accumulating instruction bloat or regressions.

Key Features

010 GitHub stars

02Pattern analysis to identify recurring issues and high-confidence fixes

03Stability Anchor (STA) conflict detection to prevent behavioral regressions

04Recursive scanning of feedback backlogs across processes, agents, and skills

05Automated changelog generation and Git-integrated commit workflows

06Consolidation-first logic that tightens existing rules rather than adding new ones

Use Cases

01Maintaining long-term documentation and version history for evolving AI systems

02Systematically upgrading agent performance based on historical execution failures

03Refining complex workflows by aggregating feedback from multiple distributed artifacts

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zoranspirkovski/pas applying-feedback

For use in Claude.ai and ChatGPT

Key Features

010 GitHub stars

02Pattern analysis to identify recurring issues and high-confidence fixes

03Stability Anchor (STA) conflict detection to prevent behavioral regressions

04Recursive scanning of feedback backlogs across processes, agents, and skills

05Automated changelog generation and Git-integrated commit workflows

06Consolidation-first logic that tightens existing rules rather than adding new ones

Use Cases

01Maintaining long-term documentation and version history for evolving AI systems

02Systematically upgrading agent performance based on historical execution failures

03Refining complex workflows by aggregating feedback from multiple distributed artifacts