zicato-step-loop

Run zicato evolution round stages individually for inspection and debugging.

4|2|Updated May 14, 2026
One-click install
npx skills add https://github.com/pedapudi/zicato --skill zicato-step-loop
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: zicato-step-loop
Source: https://github.com/pedapudi/zicato/tree/main/skills/zicato-step-loop
Command: npx skills add https://github.com/pedapudi/zicato --skill zicato-step-loop

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you inspect and debug zicato’s multi-stage evolution loop without running the entire process as an opaque operation.

Core Features & Use Cases

  • Stage-by-Stage Control: Manually run telemetry analysis, experiment proposal, candidate inspection, and tournament evaluation one step at a time.
  • Artifact Inspection: Review insights, experiments, patches, run telemetry, round logs, and journal entries between stages.
  • Safe Debugging Guardrails: Distinguish read-only inspection from LLM-backed operations that consume budget, and verify the current CLI surface before scripting.
  • Use Case: When a zicato evolution round produces an unexpected patch or promotion decision, use this Skill to inspect each artifact and identify where the loop diverged.

Quick Start

Use this Skill to manually inspect and debug one zicato evolution round from telemetry analysis through tournament evaluation.

Frequently Asked Questions about zicato-step-loop

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

FAQPage Schema
How do I debug an unexpected patch decision in an evolution loop?

Inspecting artifacts individually helps you debug unexpected promotion decisions by reviewing insights, experiments, patches, run telemetry, round logs, and journal entries between stages. This manual approach identifies the exact point where the tournament evaluation diverged from expectations.

Can I run telemetry analysis and experiment proposals separately in a CLI environment?

Yes, you can run telemetry analysis and experiment proposals separately in a CLI environment. This Skill enables manual stage-by-stage execution, allowing you to trigger telemetry analysis, experiment proposals, candidate inspection, and tournament evaluation independently.

What is the best way to review experiment patches and tournament evaluation logs?

The best way to review experiment patches and tournament logs is through artifact inspection between stages. You can manually review insights, generated patches, run telemetry, round logs, and journal entries to trace the decision path before proceeding to the next operation.

Do I need explicit approval before invoking LLM-backed operations during CLI troubleshooting?

Yes, you need explicit approval before invoking LLM-backed operations during CLI troubleshooting. Safe debugging guardrails distinguish read-only artifact inspection from budget-consuming LLM proposals and tournament operations, requiring your confirmation before execution.

Why does my evolution round produce an unexpected promotion decision?

An evolution round produces an unexpected promotion decision when a specific stage diverges. By manually inspecting the telemetry analysis, experiment proposals, and tournament evaluation artifacts one step at a time, you can identify the exact source of the divergence.