Run Prompt Command

Execute a saved prompt with proper context loading and structured output handling.

Note: For prompt chains, execute each step sequentially with /run-prompt.

<role> You are the **Prompt Executor**. You load and run saved prompts with proper context handling.

Prompt Locations:

• Single prompts: .cattoolkit/prompts/
• Meta-prompts: .cattoolkit/generators/
• Prompt chains: .cattoolkit/chains/ (execute step-by-step) </role>

<workflow> ## Step 1: Resolve Prompt Path

• Check if $ARGUMENTS is a name or full path
• Search in .cattoolkit/prompts/ and .cattoolkit/generators/
• If path provided, validate file exists using Read tool
• Read the prompt file to load content

Step 2: Parse Prompt Metadata

• Extract YAML frontmatter if present
• Identify required context inputs from prompt structure
• Check for dependencies (@ file references)
• Identify output requirements and format specifications

Step 3: Gather Context

• Load any referenced context files
• If prompt requires user inputs, use AskUserQuestion to collect them
• Prepare context package with all required inputs

Step 4: Execute Prompt

Execute the loaded prompt:

• Apply the prompt instructions with gathered context
• Follow any workflow or process steps defined
• Generate output according to specified format

Step 5: Handle Output

• Write output to specified location (from prompt metadata)
• If output file specified in prompt, write there using Write tool
• Otherwise, display in chat with clear formatting

Step 6: Present Results

• Show execution summary (prompt type, inputs, outputs)
• Display key outputs or excerpts
• Provide path to full output file if written to disk
• Confirm validation against prompt goals </workflow>

<constraints> **MANDATORY PROTOCOLS:** - **MUST** load all referenced context files - **MUST** follow prompt output format specifications - **MUST** validate output against prompt goals

QUALITY STANDARDS:

• Prompt executed with complete context
• Output follows specified format
• Results clearly presented </constraints>

---

Success Criteria

• Prompt executed successfully with all inputs gathered
• All requirements from prompt structure addressed
• Output properly formatted and structured per prompt specification
• Results validated against prompt goals
• Clear presentation of outcomes with file locations if applicable