Untitled

{Skill description -- one line explaining what this skill does}

allowed-tools: Read, Write, Edit, Glob, Grep, Bash model: {haiku/sonnet/opus}

Input

Usage: /{command-name} {arguments}

Arguments:

Argument Required Description
{arg-1} Yes {Description -- e.g., "Path to the input file or request folder"}
{arg-2} No {Description -- e.g., "Output format override (default: markdown)"}

Path Resolution

  • If an absolute path is provided, use it directly
  • If a relative path is provided, resolve from the workspace root
  • If only a name/identifier is provided, look in {default-input-directory}/
  • {Additional path resolution rules}

Input Validation

Before processing, verify:

  • Input file/folder exists at the resolved path
  • Required fields/sections are present: {list required fields}
  • {Additional validation -- e.g., "Input is in the expected language"}

If validation fails, report what is missing and stop. Do not produce partial output.

Processing Steps

  1. Read input -- Load {input-description} from the resolved path
  2. Load references -- Read relevant reference materials from content-in/{subdirectory}/:
    • {reference-file-1} -- {why it is needed}
    • {reference-file-2} -- {why it is needed}
  3. {Processing step 3} -- {Description of what happens}
  4. {Processing step 4} -- {Description of what happens}
  5. {Processing step 5} -- {Description of what happens}
  6. Apply conventions -- Ensure all naming conventions match CLAUDE.md Section 6
  7. Format output -- Structure the result using the output template at .claude/commands/templates/{template-name}

{Add or remove steps as needed. Each step should be a clear, actionable instruction.}

Output

File name pattern: {output-naming-pattern} (e.g., {concrete-example})

Format: {Markdown / Draw.io XML / JSON / other}

Destination: {output-directory}/{sub-path}/

Output Structure

The output file must contain these sections:

  1. {Section 1 -- e.g., "Header with metadata (date, author, request ID)"}
  2. {Section 2 -- e.g., "Executive summary"}
  3. {Section 3 -- e.g., "Detailed analysis"}
  4. {Section 4 -- e.g., "Recommendations"}
  5. {Section 5 -- e.g., "References and sources"}

Quality Checks

Before writing the output file, verify:

  • All required output sections are present and non-empty
  • Naming conventions match CLAUDE.md Section 6
  • Language is correct throughout ({language})
  • No placeholder text remains (no {curly-brace} markers in final output)
  • File name follows the specified pattern
  • Output is saved to the correct destination directory
  • {Domain-specific check -- e.g., "All referenced artifacts exist in the workspace"}
  • {Domain-specific check -- e.g., "Methodology terms match the loaded framework"}

Model Recommendation

Recommended: {haiku/sonnet/opus}

Rationale: {Why this model -- e.g., "This skill requires analytical reasoning and structured writing, making Sonnet the best cost/quality balance. Haiku would miss nuances; Opus is unnecessary for this complexity level."}

{If this skill can be safely downgraded after initial validation, note: "After validating output quality over 5+ runs, consider downgrading to {lower-model}."}