AgentSkillsCN

issue-creation-review

依据原始需求,全面核查 Trellis 问题的完整性、正确性与适当范围。适用于被要求“核实问题”、“验证 Trellis 问题”、“检查问题完整性”、“评审已创建的问题”时使用。

SKILL.md
--- frontmatter
name: issue-creation-review
description: Verifies Trellis issues against original requirements for completeness, correctness, and appropriate scope. Use when asked to "verify issue", "validate trellis issue", "check issue completeness", or "review created issue".
context: fork
agent: general-purpose
allowed-tools:
  - Glob
  - Grep
  - LS
  - Read
  - WebFetch
  - WebSearch
  - TodoWrite
  - mcp__perplexity-ask__perplexity_ask
  - mcp__task-trellis__get_issue
  - mcp__task-trellis__list_issues

Issue Creation Review

Verify that a created Trellis issue accurately reflects original requirements without over-engineering or missing critical elements.

Required Inputs

  • Original Requirements: The initial request or specifications
  • Created Issue: The issue ID or full issue details
  • Additional Context (optional): Clarifications or decisions made during creation

Handling Missing Information

This skill runs as a sub-agent and cannot ask questions directly. If required inputs are missing or unclear, you must return a structured response requesting clarification instead of proceeding with assumptions.

When information is missing or ambiguous, return the following structure:

code
## Clarification Needed

### Questions
1. [Specific question about missing/unclear information]
2. [Additional questions as needed]

### Context Collected So Far
- [Summary of what you've already determined]
- [Relevant codebase findings]
- [Partial analysis completed]

### Instructions for Caller
1. Gather answers to the questions above from the user
2. Re-invoke this skill with the original inputs plus the following additional context:
   - Answers to questions: [list the questions by number]
   - Previously collected context: [reference this section]

Do not make assumptions about requirements, scope decisions, or implementation details when critical information is missing.

Verification Process

1. Research Codebase Context

Before evaluating, investigate the existing system:

  • Search for similar implementations to verify consistency
  • Check architectural patterns used in the codebase
  • Identify existing utilities/libraries that should be leveraged
  • Verify integration points mentioned are valid

2. Completeness Check

Verify all required elements are present.

Common to all issue types:

  • All functional requirements from input are addressed
  • Acceptance criteria are measurable and complete
  • Dependencies/integration points are identified

Type-specific additions:

TypeAdditional Requirements
ProjectTechnical architecture specified
EpicClear scope boundaries, logical feature grouping
FeatureSpecific user-facing capability, feature integration
TaskImplementable scope, clear technical specifications

3. Correctness Check

  • Technical Accuracy: Proposed solutions align with codebase patterns
  • Requirement Alignment: Interpretation matches user intent
  • Feasibility: Approach is technically viable
  • Consistency: Aligns with existing system architecture

4. Scope Assessment

Evaluate for over-engineering:

  • Identify additions beyond the original request
  • Flag unnecessary complexity or premature optimization
  • Ensure abstractions are justified by actual requirements

Exception: Expanded scope is acceptable if explicitly requested (e.g., "comprehensive" or "future-proofed" solution).

Output

Provide a verification report covering:

  1. Issue Details: Type, ID, title
  2. Completeness: Complete/Partial/Incomplete with specific gaps
  3. Correctness: Correct/Issues Found with specific findings and codebase alignment
  4. Scope: Appropriate/Over-engineered with analysis of what was requested vs. created
  5. Recommendations: Critical issues and suggested improvements
  6. Verdict: APPROVED / NEEDS REVISION / REJECTED with summary

Use codebase evidence to support findings. Flag over-engineering only when it adds complexity without benefit.