Prism Stack Agent Skill

prism-stack — Framework Best Practices

Follow the output format defined in docs/output-kit.md — 40-line CLI max, box-drawing skeleton, unified severity indicators, compressed prose.

When to use

User asks about framework-specific patterns, component architecture, or stack guidelines.

Workflow

• Detect stack from project files (package.json, imports, config files)

   grep -r "\"react\"\|\"vue\"\|\"svelte\"\|\"next\"\|\"nuxt\"\|\"astro\"" package.json 2>/dev/null | head -5

• Search stack knowledge base:

   python3 -m prism_agent.uiux search --domain stacks --query "{stack_name}" --limit 5

• Cross-reference version — confirm guidelines match the detected major version
• Output framework-specific guidelines with code examples

Output format

┌─ Stack Guidelines — {stack_name} {version} ─────────────────────────┐
│ Category         │ Guideline                          │ Severity      │
├──────────────────┼────────────────────────────────────┼───────────────┤
│ {category}       │ {guideline}                        │ Critical      │
│ {category}       │ {guideline}                        │ High          │
│ {category}       │ {guideline}                        │ Medium        │
└──────────────────┴────────────────────────────────────┴───────────────┘

Code example:
{code_block}

Anti-patterns

• Never apply guidelines from the wrong framework version (e.g., Vue 2 patterns on Vue 3)
• Never mix framework idioms (e.g., React hooks inside Vue components)
• Never skip version detection — always confirm before outputting guidelines
• Never output framework-agnostic advice when stack-specific guidance is available

Delivery

If output exceeds the 40-line CLI budget, invoke /atlas-report with the full findings. The HTML report is the output. CLI is the receipt — box header, one-line verdict, top 3 findings, and the report path. Never dump analysis to CLI.