evaluate-findings

Classify external feedback by confidence and verify findings adversarially.

399|31|Updated Mar 11, 2026

Search Tags:#quality assurance #code review #validation #triage #feedback analysis #ai suggestions

Ready to run

npx skills add https://github.com/tobihagemann/turbo --skill evaluate-findings

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: evaluate-findings
Source: https://github.com/tobihagemann/turbo/tree/main/skills/evaluate-findings
Command: npx skills add https://github.com/tobihagemann/turbo --skill evaluate-findings

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you systematically assess and prioritize external feedback on your code or work, ensuring that only valid and actionable suggestions are considered.

Core Features & Use Cases

Automated Triage: Classifies feedback (code reviews, AI suggestions) based on confidence levels (High, Medium, Low).
Adversarial Verification: Spawns a sub-agent to challenge non-trivial findings using research tools.
Use Case: After receiving a code review with numerous comments, use this Skill to automatically filter out noise, confirm critical issues, and present a clear, actionable list of findings for your review.

Quick Start

Use the evaluate-findings skill to assess the review comments on the latest code changes.

Frequently Asked Questions about evaluate-findings

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

FAQPage Schema

How do I triage and validate code review comments automatically?▼

To triage code review comments automatically, this Skill classifies external feedback into confidence levels and uses adversarial verification to challenge non-trivial findings, returning categorized results for further action.

What is the best way to filter out noise from AI code suggestions?▼

Filtering noise from AI code suggestions involves critically assessing their applicability using a confidence-based framework, which identifies valid, actionable feedback while discarding low-confidence recommendations.

Does this feedback validation tool apply fixes to the code automatically?▼

No, this feedback validation tool does not apply fixes automatically; it strictly triages and classifies findings by confidence level, returning categorized results for the agent to act upon.

How does adversarial verification work for code review findings?▼

Adversarial verification for code review findings works by spawning a sub-agent to research and challenge medium-to-high confidence findings, ensuring only robust and valid suggestions pass through.

Can I use this Skill to assess feedback without a sub-agent?▼

No, you cannot fully assess feedback without a sub-agent, as the Skill explicitly requires a sub-agent to perform research-based challenges on medium-to-high confidence findings during adversarial verification.

Why should I classify code review feedback by confidence levels?▼

Classifying code review feedback by confidence levels helps systematically prioritize external suggestions, ensuring that only valid and actionable findings are considered while filtering out irrelevant noise.