Groundwork Help

Lists all available skills and agents provided by the Groundwork plugin. Every Groundwork capability is a skill, invoked either as /groundwork:<name> or automatically by the model.

Workflow

Step 1: Gather Component Information

Read the plugin manifest and scan directories:

1. Read ${CLAUDE_PLUGIN_ROOT}/.claude-plugin/plugin.json for overview
2. Scan ${CLAUDE_PLUGIN_ROOT}/skills/ for available skills (each <name>/SKILL.md)
3. Scan ${CLAUDE_PLUGIN_ROOT}/agents/ for available verification/research agents

Step 2: Extract Descriptions

For each skill, extract the name and description from the SKILL.md YAML frontmatter. Note its invocation tier from the frontmatter:

• disable-model-invocation: true → user-only slash workflow
• user-invocable: false → hidden library skill (model/other-skill use only)
• neither → dual (user slash + model auto)

Step 3: Group and Present

Present the skills grouped by purpose. Hide user-invocable: false library skills from the user-facing list (they are internal). Example grouping:

User-invocable skills (/groundwork:<name>)

List skills organized by purpose:

Spec-Driven Development:

• /groundwork:design-product - Define product requirements
• /groundwork:design-architecture - Design technical approach
• /groundwork:create-tasks - Generate implementation tasks

Task Execution:

• /groundwork:work-on [N] - Work on a specific task by number
• /groundwork:work-on-next-task - Work on the next task
• /groundwork:plan-task [N or description] - Plan a task or feature without implementing
• /groundwork:implement-task [N or plan-path] - Implement a previously planned task
• /groundwork:validate - Validate and autonomously fix the task worktree
• /groundwork:finalize-task [N] - Commit, merge, and clean up a validated task worktree

Quick Development:

• /groundwork:build-unplanned [description] - Build a feature from description with worktree isolation and TDD

Sync & Alignment:

• /groundwork:source-product-specs-from-code - Sync specs with code changes
• /groundwork:source-architecture-from-code - Update architecture docs
• /groundwork:check-specs-alignment - Verify spec alignment (full audit)
• /groundwork:split-specs - Convert single-file PRD into directory format
• /groundwork:split-architecture - Convert single-file architecture doc into directory format

Plugin Management:

• /groundwork:groundwork-check - Validate plugin health
• /groundwork:groundwork-help - Show this help
• /groundwork:skills - List available skills

Debugging:

• /groundwork:debug [description] - Systematic root-cause debugging
• /groundwork:swarm-debug [description] - Parallel adversarial hypothesis investigation

Design & UX:

• /groundwork:ux-design - Establish a design system

Hidden library skills

Some skills (e.g. test-driven-development, understanding-feature-requests, use-git-worktree, using-groundwork, task-validation-loop) are user-invocable: false: they are not on the slash menu and are invoked only by other skills or by the model. Use /groundwork:skills to see the full list of skills.

Step 4: Usage Examples

Provide usage examples:

/groundwork:design-product          # Start defining a new product
/groundwork:design-architecture     # Design the technical approach
/groundwork:create-tasks            # Generate implementation tasks
/groundwork:work-on-next-task       # Work on the next task
/groundwork:groundwork-check        # Check plugin health

Step 5: Additional Resources

Point users to:

• README.md in the plugin root for full documentation
• GitHub issues for support