AgentSkillsCN

netninja-repo-audit

深度审计NET-NiNJA仓库，梳理架构、入口点、废弃代码、缺失连线、破损导入、重复逻辑、待办事项、占位符处理器，以及未关联后端操作的GUI元素。当被要求审查仓库结构，或需要生成审计报告、附上文件/行号参考、绘制依赖关系图，或列出已知入口点时，可使用此技能。

SKILL.md
--- frontmatter
name: netninja-repo-audit
description: Deep audit of the NET-NiNJA repository to map architecture, entrypoints, dead code, missing wiring, broken imports, duplicated logic, TODOs, placeholder handlers, and GUI elements without backend actions. Use when asked to review the repo structure or produce an audit report, punch list with file/line refs, dependency graph, or known entrypoints list.

Netninja Repo Audit

Goal

Produce a written audit with a punch list tied to file/line, a dependency graph, and a known-entrypoints list.

Workflow

  1. •Read project docs that explain intent and current state: ARCHITECTURE.md, DECISIONS.MD, GUI_MIGRATION.md, TASKS.md, TESTING_CHECKLIST.md, WINDOWS_FEATURES.md, WSL_* docs when relevant.
  2. •Identify entrypoints and launch surfaces: main.py, netreaper_gui.py, netreaper_gui_legacy.py, netreaper_gui_windows.py, launch_gui.*, CLI scripts, and any FastAPI modules under modules/ or lib/.
  3. •Build a dependency map by walking imports between modules/, providers/, lib/, and top-level scripts. Note dynamic imports and conditional branches that affect runtime wiring.
  4. •Locate dead code and placeholders: scan for TODO, FIXME, pass, NotImplementedError, unused handlers, and UI elements without signal connections.
  5. •Cross-check GUI wiring: map clicked.connect, currentChanged, and other signal hookups to handlers, then trace handlers to backend calls (job pipeline, providers, modules). Flag no-ops and missing outputs.
  6. •Validate import integrity: flag circular or broken imports, unused modules, duplicated logic, and inconsistent naming that hides bugs.

Output Format

  • •Audit summary with major risks and themes
  • •Punch list: path:line with short fix guidance
  • •Dependency graph: adjacency list or bullet map
  • •Known entrypoints: executable files, GUI entry, API entry, tests

Tips

  • •Prefer rg with targeted patterns, then open files for context.
  • •Keep findings grounded in code references with exact file/line.