review-iteration

Community

Read-only diagnostics for pipeline results

Authorestebandegetau
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill provides a structured, read-only analysis of pipeline run results to help users understand failures, map recurring error patterns, and identify next investigative steps without modifying any files or code.

Core Features & Use Cases

  • Structured Diagnosis: Aggregates pipeline outputs, codebook prompts, iteration logs, and success criteria to produce a concise pass/fail summary and root-cause hypotheses.
  • Stage-specific Analysis: Performs tailored analyses for S1 behavioral tests, S2 zero-shot evaluation, and S3 error/ablation studies with example deep-dives and confusion patterns.
  • Investigation Guidance: Ranks ablations and suggests high-leverage follow-ups so users can prioritize experiments or codebook edits.
  • Use Case: After running the targets pipeline, run this Skill to determine which tests failed, why they failed, and what to investigate next before recording changes.

Quick Start

Ask the Skill to analyze codebook c1 at stage s2 and summarize failing metrics, dominant error patterns, and ranked next investigations.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: review-iteration
Download link: https://github.com/estebandegetau/Fiscal-shocks/archive/main.zip#review-iteration

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 223,000+ vetted skills library on demand.