tpl-agentic-dev-base/.claude/commands/docs/parse-prd.md

allowed-tools, description

allowed-tools	description
TodoWrite, mcp__taskmaster-ai__parse_prd, mcp__taskmaster-ai__add_tag, mcp__taskmaster-ai__use_tag, mcp__taskmaster-ai__list_tags, mcp__taskmaster-ai__get_tasks	Parse a PRD into Task Master tasks with optional tag creation

Parse PRD into Task Master Tasks

Context

• User Request: $ARGUMENTS
• Current directory: !pwd
• Task Master state: !cat .taskmaster/state.json 2>/dev/null || echo "No state file yet"
• Current tag: !jq -r '.currentTag // "master"' .taskmaster/state.json 2>/dev/null || echo "master"
• Available tags: !jq -r '.tags | keys | join(", ")' .taskmaster/tasks/tasks.json 2>/dev/null || echo "No tags yet"
• PRD files: !ls -la .taskmaster/docs/prd*.md 2>/dev/null | tail -5 || echo "No PRD files found"

Goal

Parse a Product Requirements Document (PRD) into structured Task Master tasks. This command handles tag creation, context switching, and PRD parsing in a streamlined workflow.

Process

1. Determine PRD Location

Think about which PRD file the user wants to parse.

Check for:

• Explicit PRD path in
• Default PRD location: .taskmaster/docs/prd.txt or .taskmaster/docs/prd.md
• Tag-specific PRD: .taskmaster/docs/prd-[tag-name].md

2. Tag Context Decision

Determine if we need a new tag:

• If PRD is for a specific feature → Create new tag
• If updating existing work → Use current tag
• If starting fresh → Consider new tag

3. Execute Parse Workflow

Based on context:

1. Create new tag if needed
2. Switch to appropriate tag
3. Parse the PRD
4. Generate tasks with proper numbering
5. Suggest next steps

Execution Steps

Scenario 1: Parse with New Tag Creation

If the user wants to parse a feature-specific PRD:

1. **Create a new tag** for this feature:
   Using: add_tag with name and description

2. **Switch to the new tag**:
   Using: use_tag to set context

3. **Parse the PRD**:
   Using: parse_prd with the PRD path

4. **Confirm success**:
   Show task count and suggest next steps

Scenario 2: Parse in Current Context

If parsing into the current tag:

1. **Confirm current tag** is appropriate
   Show current tag context

2. **Parse the PRD directly**:
   Using: parse_prd with the PRD path

3. **Show results**:
   Display generated tasks summary

Scenario 3: Parse Default PRD

If no specific PRD mentioned:

1. **Check for default PRD**:
   Look for .taskmaster/docs/prd.txt or prd.md

2. **Confirm with user** if found
3. **Parse the default PRD**:
   Using: parse_prd

Interactive Flow

Based on User Request, determine the appropriate flow:

If arguments include a tag name:

1. Create the tag
2. Switch to it
3. Parse the corresponding PRD

If arguments include a PRD path:

1. Ask if a new tag is needed
2. Parse the specified PRD

If no arguments:

1. Check current tag context
2. Look for default PRD
3. Proceed with parsing

Best Practices

DO:

• Check tag context before parsing
• Use descriptive tag names for features
• Keep PRDs organized by feature/tag
• Verify PRD exists before parsing
• Show task summary after parsing

DON'T:

• Parse into master tag for feature work
• Overwrite existing tasks without confirmation
• Mix unrelated features in one tag
• Skip tag creation for new features

Example Usage

# Parse default PRD in current context
/project:parse

# Parse specific PRD with new tag
/project:parse user-auth feature

# Parse existing PRD file
/project:parse .taskmaster/docs/prd-payments.md

Natural Language Examples

Since MCP supports natural language:

"Please parse my PRD for the user authentication feature"
"Create tasks from the payments PRD and put them in a new tag"
"Parse the default PRD into the current tag context"

Next Steps

After parsing, suggest:

1. View generated tasks: Use /next to see the first task
2. Analyze complexity: Run complexity analysis if many tasks
3. Expand tasks: Break down complex tasks into subtasks
4. Start implementation: Begin with the highest priority task

Task Tracking

Add parsed PRD to todo list for tracking:

{
  content: "Parse PRD: [filename]",
  status: "completed",
  priority: "high"
}

This helps track which PRDs have been processed and when.