用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/Texarkanine/inquirerjs-checkbox-search --skill niko-qa命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Niko Memory Bank System - Preflight Phase - Pre-Build Plan Validation
Niko Memory Bank System - Niko Phase - Initialization & Entry Point
Guidelines for using Mermaid diagrams to increase comprehension of complex processes. Apply when planning non-trivial work involving multiple components, interactions, data models, or process flows, or when describing a system with complex interactions where prose alone becomes unwieldy.
基于 SOC 职业分类
正在显示 SKILL.md
| name | niko-qa |
| description | Niko Memory Bank System - QA Phase - Post-Implementation Semantic Review |
This command performs a structured semantic review of the code just implemented against the original plan. It catches over-engineering, incomplete implementations, pattern violations, and implementation debris that mechanical checks (lint/build/test) cannot detect.
Read:
memory-bank/active/tasks.mdmemory-bank/active/projectbrief.mdmemory-bank/systemPatterns.mdmemory-bank/active/activeContext.mdmemory-bank/active/progress.mdmemory-bank/active/creative/Verify Prerequisites
memory-bank/active/tasks.md for build phase completionReview the code just implemented against the original plan and apply these constraints:
KISS: Simplify over-engineered logic; flatten unnecessary abstractions or indirection layers introduced during the build. If a simpler construct achieves the same outcome, prefer it. Do not preserve complexity merely because it was part of the initial implementation approach.
DRY: Consolidate any duplicate code, boilerplate, or redundant patterns introduced during iterative development into clean, reusable constructs. Cross-reference new code against existing utilities and helpers to avoid reinventing what the codebase already provides.
YAGNI: Prune speculative code, "just-in-case" variables, unused parameters, and features not explicitly required by the plan. If it wasn't asked for, it doesn't ship.
Completeness: Verify every requirement from the original plan was actually implemented - not stubbed, TODO'd, commented-as-pseudocode, or hand-waved. Treat any // TODO or placeholder value introduced during this session as a blocking deficiency, not a future suggestion.
Regression: Confirm no existing architectural patterns were broken - naming conventions, casing, error handling strategies, import styles, file structure, and established abstractions must remain consistent across all affected projects. New code must be indistinguishable in style from surrounding code and integrate as a natural extension of existing architecture, not an accretion layer.
Integrity: Replace any hardcoded shortcuts, magic numbers, placeholder strings, or debug artifacts (console.log, print("HERE")) introduced as temporary scaffolding. If it was a means to an end during development, it does not survive into the final commit.
Documentation: Verify that any project documentation (README files, doc comments, memory bank persistent files, configuration docs, user-facing guides) affected by the code changes was updated alongside those changes. Treat missing documentation updates as an incomplete implementation — same severity as a missing requirement.
Apply Fixes or Fail
Generate QA Report
memory-bank/active/.qa-validation-statusmemory-bank/active/tasks.md with QA resultsHandle Results
🚨 Printing this notice is NOT the end of this phase. After printing, continue immediately to the next step - do not stop.
Update memory-bank/active/progress.md to record completion of the QA phase.
When QA review is complete, print:
# QA Result
✅ PASS
1. **Findings** - bulleted list of each semantic finding and the fix applied (or why it blocks)
# QA Result
❌ FAIL
1. **Findings** - bulleted list of each semantic finding and the fix applied (or why it blocks)
## Next Steps
(the next command, if any, based on the current complexity-level's workflow & QA result)