Quality Stack

Scan a project's build configuration across JVM, Node.js, and Python ecosystems, cross-reference against curated research documents, and assist with tool setup.

Pre-flight

1. Run the orchestrator — it auto-detects ecosystems:
   bashCopy

   python3 <skill-path>/scripts/scan_project.py <project-root>
   

2. If "error": "no_ecosystem_detected", check nearby_project_files for subproject paths.
3. Monorepo? — use --recursive or --ecosystem to force a specific scanner:
   bashCopy

   python3 <skill-path>/scripts/scan_project.py --recursive <project-root>
   python3 <skill-path>/scripts/scan_project.py --ecosystem node <project-root>
   

4. Legacy (JVM only) — scan_tooling.py still works as a backwards-compatible wrapper.

Two-Phase Workflow

Phase 1: Recommend

1. Run the scanner on the project root (see Pre-flight above).

2. Fetch research documents via WebFetch — only for detected ecosystems:

   JVM (when ecosystems contains "jvm"):

   Copy

   https://raw.githubusercontent.com/joaquimscosta/arkhe-claude-plugins/main/docs/research/jvm-quality-tools-evaluation.md
   https://raw.githubusercontent.com/joaquimscosta/arkhe-claude-plugins/main/docs/research/kotlin-spring-boot-testing-ecosystem.md
   

   Node.js (when ecosystems contains "node"):

   Copy

   https://raw.githubusercontent.com/joaquimscosta/arkhe-claude-plugins/main/docs/research/node-quality-tools-evaluation.md
   

   Python (when ecosystems contains "python"):

   Copy

   https://raw.githubusercontent.com/joaquimscosta/arkhe-claude-plugins/main/docs/research/python-quality-tools-evaluation.md
   

   Cross-cutting (always):

   Copy

   https://raw.githubusercontent.com/joaquimscosta/arkhe-claude-plugins/main/docs/research/cross-cutting-devtools-evaluation.md
   

   If WebFetch fails, warn the user and proceed using scanner results + LLM knowledge only.

3. Cross-reference scanner output against research recommendations per ecosystem:

   • Identify tools recommended but missing from the project
   • Check status field: disabled or config-only tools need attention
   • Flag outdated or superseded tools
   • Apply ecosystem-specific SKIP rules (see WORKFLOW.md)
   • Review tool_config for threshold values and settings

4. Generate the recommendation report using the format in WORKFLOW.md.

Phase 2: Setup

After presenting the report, present tools for selection using the multi-round protocol in WORKFLOW.md. Group by ecosystem and priority tier. Include effort estimates. After user completes selection across all rounds:

1. Read the relevant research doc section for setup instructions
2. Apply Setup Guards — resolve versions, check compatibility
3. For each selected tool, apply changes per ecosystem:
   • JVM: Add Gradle/Maven plugin, test deps, config files
   • Node.js: pnpm add -D, tsconfig edits, config file creation
   • Python: uv add --dev, pyproject.toml edits, config file creation
   • Cross-cutting: CI/CD workflow steps, Lefthook config, EditorConfig
4. Verify each tool after configuration — run the tool's check command, verify filter patterns against actual codebase paths, and check for config inheritance conflicts. See WORKFLOW.md Post-Setup Verification.
5. Re-run the scanner to confirm all tools detected

Priority Classification

Ecosystem-aware rules — see WORKFLOW.md for full classification tables per ecosystem.

Key rules:

• JVM Pure Kotlin: SKIP Error Prone, SpotBugs; JVM Pure Java: SKIP Detekt, ktlint, MockK
• JVM Spring Boot 4+: SKIP REST Assured, NOW MockMvcTester
• Node.js no linter: NOW ESLint; no formatter + no Biome: NOW Prettier
• Node.js TypeScript not strict: NOW enable strict
• Python no linter: NOW Ruff; no type checker + has type annotations: NOW mypy

Cross-Cutting Tools

Research Documents

Fetch via WebFetch at runtime — only for detected ecosystems:

• JVM Quality Tools: jvm-quality-tools-evaluation.md
• JVM Testing Ecosystem: kotlin-spring-boot-testing-ecosystem.md
• Node.js Quality Tools: node-quality-tools-evaluation.md
• Python Quality Tools: python-quality-tools-evaluation.md
• Cross-Cutting Tools: cross-cutting-devtools-evaluation.md

Scanner Architecture

scripts/
  scan_project.py          # Orchestrator — auto-detects + merges
  scan_jvm.py              # JVM scanner (Gradle/Maven)
  scan_node.py             # Node.js/TypeScript scanner
  scan_python.py           # Python scanner
  scan_cross_cutting.py    # Cross-cutting tools (CI, hooks, security)
  shared.py                # Shared utilities
  scan_tooling.py          # Legacy wrapper → scan_jvm.py

References

• Workflow: See WORKFLOW.md for classification rules and report format
• Examples: See EXAMPLES.md for realistic audit scenarios
• Troubleshooting: See TROUBLESHOOTING.md for scanner issues