skill-auto-evolver

What problem does it solve?

It eliminates manual, ad-hoc performance debugging for Skills by collecting execution telemetry, identifying bottlenecks and error patterns, and producing prioritized optimization actions so teams can improve reliability and latency with data.

Core Features & Use Cases

• Execution Tracing & Collection: Non-intrusive decorators and a lightweight SQLite store capture start/end timestamps, durations, success/failure and optional inputs.
• Performance Analysis: Compute P50/P95/P99 latencies, detect latency spikes, and surface resource-intensive execution paths.
• Optimization & Experimentation: Generate data-driven suggestions, create optimized versions, run A/B experiments, and manage regression test baselines.
• Use Case: Collect telemetry for a slow skill, analyze tail latency and error patterns, generate an optimization plan, and validate improvements with an A/B test and regression suite.

Quick Start

Start collecting executions for a skill, run the analyzer to identify bottlenecks, and request optimization suggestions in one workflow using the skill-auto-evolver commands.