How does it collect performance data?

The skill integrates with Loki to query enhanced telemetry logs, tracking event types such as tool results, session ends, and error patterns.

Which tools are required to use this skill?

It is designed to work within the Skrillz ecosystem, integrating with observability-analyzer, skill-updater, and enhanced-telemetry hooks.

What is the primary benefit of the Skill Optimizer?

It creates a continuous feedback loop that automatically identifies and fixes weaknesses in Claude's specialized behaviors using real usage data and telemetry.

Is it safe to let this skill update my code automatically?

The skill includes a safety classification system. Low-risk changes like adding validation or improving error messages can be auto-applied, while core logic changes always require manual review.

Does it provide metrics on how much a skill has improved?

Yes, it includes effectiveness tracking that compares error rates and success percentages before and after an improvement is deployed.

Observability-Driven Skill Optimizer

Name: Observability-Driven Skill Optimizer
Author: adaptationio

byadaptationio

0•

Analytics & Monitoring

Automates the continuous improvement of Claude Code skills by analyzing telemetry patterns and applying performance-enhancing updates.

This skill implements a sophisticated self-improvement loop for Claude Code by bridging the gap between runtime observability and skill refinement. It queries enhanced telemetry logs from Loki to identify recurring failure patterns, inefficient tool sequences, and context management issues. By analyzing real-world usage data, it generates actionable recommendations or automated updates to existing skills—such as adding file existence checks or optimizing redundant tool calls—ensuring that Claude's specialized capabilities evolve to become more reliable and efficient over time.

Key Features

01Automated pattern detection for tool failures and workflow inefficiencies

02Safety-first classification for manual or automated code changes

03Real-time telemetry analysis using advanced LogQL queries

04Performance tracking to measure the success of deployed updates

050 GitHub stars

06Generation of data-backed skill improvement recommendations

Use Cases

01Reducing operation latency by caching results of redundant file discovery patterns

02Eliminating recurring bash errors by automatically adding environment existence checks

03Optimizing token consumption by identifying triggers for high-frequency context compaction

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adaptationio/skrillz skill-improvement-from-observability

For use in Claude.ai and ChatGPT

Download Skill