Agent Skills: agent-creator

  <substep n="1a" title="Track Progress">
    <action if="first iteration">Create todo list from all instruction step titles (TodoWrite)</action>
    <action>Mark current step as in_progress (TodoWrite)</action>
  </substep>

  <substep n="1b" title="Handle Step Attributes">
    <check>If optional="true" and NOT YOLO -> Use Question tool to ask user whether to include (Yes/Skip)</check>
    <check>If if="condition" -> Evaluate condition</check>
    <check>If for-each="item" -> Repeat step for each item</check>
    <check>If repeat="n" -> Repeat step n times</check>
  </substep>

  <substep n="1c" title="Execute Step Content">
    <action>Process step instructions (markdown or XML tags)</action>
    <action>Replace {{variables}} with values (ask user if unknown)</action>
    <execute-tags>
      <tag>action -> Perform the action</tag>
      <tag>action type="script" -> Execute as shell command, capture output</tag>
      <tag>check if="condition" -> Conditional block wrapping actions (requires closing &lt;/check&gt;)</tag>
      <tag>ask -> Prompt user using Question tool and WAIT for response. When options exist, present them as selectable choices. For open-ended questions, provide helpful suggested options with custom input enabled</tag>
      <tag>invoke-skill name="skill-name" -> Execute another skill via OpenCode skill system</tag>
      <tag>invoke-agent name="subagent" prompt="task" -> Delegate to subagent via Task tool</tag>
      <tag>goto step="x" -> Jump to specified step (used for explicit loop-back in Q-R style flows)</tag>
    </execute-tags>
  </substep>

  <substep n="1d" title="Handle template-output Tags">
    <if tag="template-output">
      <mandate>Generate content for this section</mandate>
      <mandate>Save to file (Write first time, Edit subsequent)</mandate>
      <action>Display generated content</action>
      <action>Use Question tool to present checkpoint options. WAIT for response.
        <options>
          <option label="Continue" description="Approve this section and move to the next step" />
          <option label="Advanced Elicitation" description="Explore this section deeper before proceeding" />
          <option label="YOLO" description="Auto-complete the rest of this workflow without stopping" />
        </options>
        <if response="Advanced Elicitation">
          <action>Invoke advanced-elicitation skill</action>
        </if>
        <if response="Continue">
          <action>Continue to next step</action>
        </if>
        <if response="YOLO">
          <action>Enter YOLO mode for the rest of the workflow</action>
        </if>
      </action>
    </if>
  </substep>

  <substep n="1e" title="Step Completion">
    <action>Mark current step as completed (TodoWrite)</action>
    <check>If no special tags and NOT YOLO:</check>
    <action>Use Question tool to ask:
      <options>
        <option label="Continue" description="Proceed to the next step" />
        <option label="Edit" description="Revise this step before moving on" />
      </options>
    </action>
  </substep>
</step>

<step n="2" title="Completion">
  <check>Confirm all instruction steps have been executed</check>
  <output>Workflow complete.</output>
</step>

<output>
  A complete agent source directory under .vera/agents/{agent_id}/
  containing agent.yaml, frontmatter.md, and agent-body.md, plus any
  additional include files the agent needs. The manifest gets updated
  with the new agent's artifact entry.
</output>

<key-principles>
  <principle>
    Templates are self-documenting — each content template under
    references/templates/ contains embedded FILL instructions that teach
    how to fill every section. The skill orchestrates the process; the
    templates own the content knowledge.
  </principle>
  <principle>
    Shared structural blocks live in dubstack's .vera/shared/ —
    activation.md, workflow-engine.md, baseline/context-md-awareness.md,
    baseline/thoughts-cli-awareness.md. New agents reference these via
    the agent.yaml `includes:` list rather than duplicating content.
  </principle>
  <principle>
    Persona principles are MENTAL POSTURE, not rule lists. The persona
    template's FILL instructions warn explicitly against rule-list drift
    (workflow constraints disguised as principles). Push back during the
    fill conversation when the user proposes bolded-name commandments —
    those belong in skill prompts or activation rules, not in the
    persona block.
  </principle>
  <principle>
    Cross-template consistency is enforced before completion: skills
    named in agent.yaml permissions must appear in the menu; menu skill
    names must match real skill IDs under .vera/skills/; subagent input
    contracts must be parseable by the orchestrator that will dispatch
    them.
  </principle>
</key-principles>

<step n="1" title="Discovery & Context">
  <action>Load and execute: references/steps/step-01-discovery.xml</action>
  <note>
    Gather requirements, determine mode (primary/subagent), identify
    skills. Check for architecture documents (workflow specs,
    implementation specs) the agent must conform to.
  </note>
</step>

<step n="2" title="Template Selection">
  <action>Load and execute: references/steps/step-02-template-selection.xml</action>
  <note>
    Auto-select structural templates by mode. Auto-select required
    content templates (persona + menu for primary; persona +
    subagent-instructions for subagent). Detect missing project
    infrastructure (shared/activation.md, shared/agent-activation.md,
    shared/agent-close.md, shared/baseline/*) and copy from bundled
    references when absent. Confirm the composition order in
    agent.yaml includes.
  </note>
</step>

<step n="3" title="Scaffold & Configure">
  <action>Load and execute: references/steps/step-03-scaffold.xml</action>
  <note>
    Gather identity (agent_id, name, title, emoji). Configure
    frontmatter (mode, color, permission map). Write agent.yaml with
    includes. For primary agents, includes typically begin with
    frontmatter.md, then shared/activation.md, then any
    shared/baseline/* fragments the agent needs, then agent-body.md.
  </note>
</step>

<step n="4" title="Fill Content Templates" critical="true">
  <action>Load and execute: references/steps/step-04-fill-process.xml</action>
  <note>
    THIS IS THE CRAFT STEP. Work through each selected content template,
    reading its embedded FILL instructions, discussing each section with
    the user, and writing quality content. The persona template's
    instructions are particularly important — they encode the BMAD
    mental-posture convention for principles.
  </note>
</step>

<step n="5" title="Validate & Complete">
  <action>Load and execute: references/steps/step-05-validate.xml</action>
  <note>
    Verify files exist with expected structure. Cross-reference
    consistency between agent.yaml permissions, menu items, and skills.
    Run `vera build` in a dry-run mode if available to surface
    compile-time issues before declaring the agent complete. Update the
    relevant build group in .vera/vera.manifest.yaml.
  </note>
</step>

<next-steps>
  <step>Create each new skill the agent depends on via [CS] Create Skill</step>
  <step>Build dubstack: `vera build`</step>
  <step>Verify the compiled artifact at the configured output path</step>
  <step>Test in OpenCode by switching to the new agent</step>
</next-steps>