skill-tuning

Diagnose and repair skill execution failures in SKILL.md directories.

76|5|Updated Jul 7, 2026
One-click install
npx skills add https://github.com/catlog22/pi-maestro-flow --skill skill-tuning-catlog22
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-tuning
Source: https://github.com/catlog22/pi-maestro-flow/tree/main/.pi/skills/skill-tuning
Command: npx skills add https://github.com/catlog22/pi-maestro-flow --skill skill-tuning-catlog22

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill finds and fixes execution problems in other skills, such as context explosion, forgotten constraints, broken data flow, fragile agent coordination, and excessive token usage.

Core Features & Use Cases

  • Structured diagnosis: Analyzes a target skill across context, memory, dataflow, agent behavior, docs, and token consumption.
  • Fix proposal and repair: Generates targeted remediation plans using strategies like sliding windows, schema enforcement, error wrapping, and prompt compression.
  • Quality-gated verification: Re-checks fixes against clear pass, review, and fail thresholds before completing the tuning session.
  • Use case: A skill starts losing its original instructions after several phases, and this Skill traces the cause, proposes a repair, and verifies that the issue is resolved.

Quick Start

Use the skill-tuning skill to diagnose the target skill and generate prioritized fixes for its execution and orchestration issues.

Frequently Asked Questions about skill-tuning

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix an AI agent losing its original instructions after several execution phases?

To fix an agent losing instructions, diagnose the target skill for context explosion and forgotten constraints, then apply remediation strategies like sliding windows or prompt compression to restore reliable execution.

Why does my multi-agent orchestration workflow have broken data flow and fragile coordination?

Fragile agent coordination and broken data flow occur when skills lack schema enforcement and error wrapping. Diagnose the orchestration skill to trace execution failures and generate targeted repair plans for agent behavior.

What is the best way to diagnose excessive token usage in a self-contained skill directory?

The best way to diagnose excessive token usage is to analyze the target skill across context, memory, dataflow, and token consumption, producing a structured analysis that identifies performance bottlenecks and prompt compression opportunities.

How do I verify that skill optimization repairs actually resolve execution failures?

To verify skill optimization repairs, re-check the applied fixes against predefined quality gates using clear pass, review, and fail thresholds before completing the tuning session to ensure the execution failure is fully resolved.

Can I use skill diagnosis to repair documentation inconsistency in a skill with a SKILL.md entry point?

Yes, skill diagnosis can repair documentation inconsistency in self-contained directories with a SKILL.md entry point by analyzing the target skill's docs and generating verification-ready fixes for reliability.

What are the limitations of applying prompt compression and sliding windows for state management?

Limitations of prompt compression and sliding windows for state management depend on the complexity of the target skill's multi-agent orchestration, requiring structured diagnosis to ensure constraints are not lost during context control.