evaluate-issue

Evaluate enriched GitHub issues to keep, complete, or split them.

11|Updated Dec 28, 2024
One-click install
npx skills add https://github.com/wadvanced/aurora_uix --skill evaluate-issue
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: evaluate-issue
Source: https://github.com/wadvanced/aurora_uix/tree/main/.claude/skills/evaluate-issue
Command: npx skills add https://github.com/wadvanced/aurora_uix --skill evaluate-issue

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill turns an already-enriched GitHub issue into a clear decision on whether the work is correctly sized for a single coding pass, should be marked completed, or needs to be split into smaller child issues.

Core Features & Use Cases

  • Completion & progress detection: Determines whether the issue is completed, partially done, or fresh by inspecting the enriched spec and review-gaps markers.
  • Remaining-work slicing: For partially completed issues, focuses evaluation only on the unticked acceptance criteria and the files they implicate, avoiding re-evaluating already-done work.
  • Evidence-backed sizing & model tier recommendation: Produces a file/spread-driven assessment and recommends a model tier for KEEP decisions, or proposes split strategies when oversized/complex.

Quick Start

Evaluate an already-enriched issue by asking the AI to run evaluate-issue on issue number N, after confirming the issue body already includes the enriched-spec marker block.

Frequently Asked Questions about evaluate-issue

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

FAQPage Schema
How do I decide whether to split or keep a GitHub issue for a single coding pass?

Evaluate an enriched GitHub issue to decide whether to keep it for a single implementation pass, mark it completed, or recommend splitting it into smaller children based on remaining scope. The assessment produces evidence-backed sizing and model tier recommendations.

How does issue evaluation handle partially completed work on GitHub?

Issue evaluation for partially completed work focuses only on unticked acceptance criteria and their implicated files. This remaining-work slicing avoids re-evaluating already done work by inspecting enriched specs and review-gaps markers to determine completion status.

What prerequisites are required before evaluating an issue for task sizing?

Evaluating an issue requires the GitHub issue body to already include an enriched-spec marker block. The Skill enforces this strict precondition and re-reads the issue via GitHub issue view and comments before applying file and layer driven tiering.

Does the issue evaluation process recommend an AI model tier for implementation?

Yes, the issue evaluation produces an evidence-backed file and spread-driven assessment that recommends a specific model tier for KEEP decisions. When an issue is oversized or complex, it proposes split strategies instead of a model recommendation.

Can I evaluate large cross-layer issues that have already been enriched?

Yes, this issue evaluation applies to scenarios where issues are already decorated with enriched-spec and optional review-gaps sections, including large, cross-layer, or partially completed work. It performs an idempotent assessment and persists the results directly to the issue body.