SKILL.md Generation.

You are an expert in writing Agent Skills for AI coding agents. Generate the YAML frontmatter block for a SKILL.md file.

SKILL_NAME: {{SKILL_NAME}}
WHAT_IT_DOES: {{WHAT_IT_DOES}}
TRIGGER_REQUESTS: {{TRIGGER_REQUESTS}}
EXCLUSIONS: {{EXCLUSIONS}}
AUTHOR: {{AUTHOR}}
TAGS: {{TAGS}}

Generate the frontmatter block following these rules:

1. name: kebab-case, unique, describes the skill precisely
2. description: the most important field — the agent reads ONLY this during the Discovery stage to decide whether to activate. Write it as follows:
   - Sentence 1: What this skill does (specific output, not vague purpose)
   - Sentence 2: "Use when [specific trigger phrases the user might type]"
   - Sentence 3: "Do NOT use for [explicit exclusions — prevents false positives]"
   - Total: 2–5 sentences. No longer. No shorter than 2.
3. version: "1.0.0"
4. metadata: include author, tags, and last-reviewed

Output the frontmatter block in a fenced code block, then explain:
- Why the description will activate correctly
- Which phrase(s) in the description are the primary trigger
- Which phrase(s) prevent false positives
- One improvement the author could make to further tighten the scope

Do not generate the full SKILL.md body — frontmatter only.

You are an expert in writing Agent Skills for AI coding agents. Your task is to improve a SKILL.md description field so it activates the skill reliably and avoids false positives.

CURRENT_DESCRIPTION:
{{CURRENT_DESCRIPTION}}

PROBLEM: {{PROBLEM}}
(e.g. "The agent activates this skill when the user asks for unit tests, which it should not handle" or "The agent never activates this skill even when the user asks to write Playwright tests")

SKILL_ACTUAL_PURPOSE: {{SKILL_ACTUAL_PURPOSE}}
SHOULD_ACTIVATE_FOR: {{SHOULD_ACTIVATE_FOR}}
SHOULD_NOT_ACTIVATE_FOR: {{SHOULD_NOT_ACTIVATE_FOR}}

Rewrite the description field following these rules:
1. Sentence 1: State the specific output this skill produces (not just the category of task)
2. Sentence 2: "Use when [specific trigger conditions or user phrases]"
3. Sentence 3: "Do NOT use for [explicit exclusions]"
4. Length: 2–5 sentences total. The agent reads ONLY this field at startup (Discovery stage).

Output:
1. The improved description (as it would appear in SKILL.md frontmatter)
2. What changed and why — specifically which phrases fix the stated problem
3. A test: 3 example user requests that should now activate the skill, and 3 that should not

Do not generate or modify any other part of the SKILL.md.

You are an expert in writing Agent Skills for AI coding agents. Convert the prompt below into a complete, well-structured SKILL.md.

EXISTING_PROMPT:
{{EXISTING_PROMPT}}

SKILL_NAME: {{SKILL_NAME}}
AUTHOR: {{AUTHOR}}
TARGET_AGENT: {{TARGET_AGENT}}

Convert the prompt into a SKILL.md following these rules:

1. Frontmatter:
   - name: kebab-case identifier
   - description: 2–5 sentences — what it does, when to use (trigger phrases), when NOT to use (exclusions). This is the activation trigger.
   - version: "1.0.0"
   - metadata: author, tags, last-reviewed

2. Instruction body (use these h2 sections):
   - "## When to use" — bullet list of specific trigger conditions
   - "## When NOT to use" — bullet list of explicit exclusions
   - "## Inputs" — what the user must provide
   - "## Instructions" — numbered steps derived from the prompt logic
   - "## Output format" — what the agent should produce
   - "## Anti-patterns" — common mistakes from the original prompt's cautions
   - "## Safety" — any safety constraints (always include: no credentials, output is a draft, review before use)

3. Identify any implicit assumptions in the original prompt and make them explicit as inputs or constraints.

Output the complete SKILL.md in a fenced code block, then note:
- Any information that was implicit in the prompt and is now made explicit
- Any missing sections (inputs, output format) that you had to infer
- One recommendation for improving the skill beyond the original prompt

You are an expert in writing Agent Skills for AI coding agents. Design the folder structure and supporting file contents for a multi-file Agent Skill.

SKILL_NAME: {{SKILL_NAME}}
WORKFLOW_DESCRIPTION: {{WORKFLOW_DESCRIPTION}}
SUPPORTING_MATERIALS: {{SUPPORTING_MATERIALS}}
(e.g. "coding conventions doc, test file template, setup script that installs dependencies")
AGENT_TOOL: {{AGENT_TOOL}}

Design the complete skill folder:

1. Show the full directory tree for the skill (as a code block).
2. For each directory (references/, scripts/, templates/, examples/):
   - State what it contains and why
   - Explain how the SKILL.md instruction body references it (exact wording)
   - State whether it should be in this directory or kept outside the skill
3. Write the SKILL.md instruction body section that references the supporting files. Use explicit language like:
   - "Before generating tests, read references/playwright-conventions.md and apply all rules."
   - "Use templates/spec-template.ts as the starting file — copy it then fill it in."
   - "Run scripts/setup.sh to install dependencies before generating code. Require user confirmation first."
4. Security guidance for scripts/:
   - What must scripts never do (embed credentials, make network calls without confirmation, write outside the project directory)
   - How to validate inputs inside scripts
   - Whether a script is necessary or if a plain reference would suffice

Output:
1. Directory tree (code block)
2. Supporting file analysis (per directory)
3. SKILL.md instruction body excerpt that references the files
4. A recommendation on which supporting files are essential vs. optional for this workflow