Installable skillMIT

intent-capture

Capture user intent through guided conversation. Exploratory phase with high degrees of freedom.

BY fabriqaaiUPDATED Oct 27, 2026
INTENDED FOR✳Claude Code>_Codex◈OpenClaw↗CursorInstall in a listed agent.

instructions.md

---
name: intent-capture
description: Capture user intent through guided conversation. Exploratory phase with high degrees of freedom.
version: 1.0.0
---

<objective>
Capture user intent through guided conversation.
</objective>

<triggers>
  - No active intent exists
  - User wants to start something new
</triggers>

<degrees_of_freedom>
  **HIGH** — This is a creative, exploratory phase. Ask open-ended questions. Don't constrain prematurely.
</degrees_of_freedom>

<llm critical="true">
  <mandate>NEVER assume requirements — ALWAYS ask clarifying questions</mandate>
  <mandate>Capture the "what" and "why" — leave the "how" for decomposition</mandate>
  <mandate>Let user describe freely — don't interrupt</mandate>
</llm>

<flow>
  <step n="1" title="Initial Question">
    <ask>What do you want to build?</ask>
    <listen>Let user describe freely. Don't interrupt.</listen>
  </step>

  <step n="2" title="Elicit Context">
    <action>Based on response, ask follow-up questions:</action>

    <question if="unclear who benefits">
      Who is this for? Who will use this feature?
    </question>

    <question if="unclear problem">
      What problem does this solve? What's painful today?
    </question>

    <question if="unclear scope">
      What's the minimum that would be valuable? What can wait?
    </question>

    <question if="unclear constraints">
      Any technical constraints? Existing systems to integrate with?
    </question>

    <question if="unclear success">
      How will you know this is working? What does success look like?
    </question>
  </step>

  <step n="3" title="Summarize Understanding">
    <output>
      Let me make sure I understand:

      **Goal**: {summarized goal}

      **Users**: {who benefits}

      **Problem**: {what pain this solves}

      **Success Criteria**:
      - {criterion 1}
      - {criterion 2}
      - {criterion 3}

      **Constraints**:
      - {constraint 1}
      - {constraint 2}

      Is this accurate? [Y/n/edit]
    </output>
    <check if="response == n or edit">
      <action>Ask specific clarifying questions</action>
      <goto step="3"/>
    </check>
  </step>

  <step n="4" title="Generate Intent Brief">
    <action>Create intent ID from title (kebab-case)</action>
    <action>Generate intent brief using template: templates/brief.md.hbs</action>
    <action>Create directory: .specs-fire/intents/{intent-id}/</action>
    <action>Save: .specs-fire/intents/{intent-id}/brief.md</action>
  </step>

  <step n="5" title="Update State">
    <action>Add intent to state.yaml</action>
    <action>Set intent status to "in_progress"</action>
  </step>

  <step n="6" title="Transition">
    <output>
      **Intent captured**: "{intent-title}"

      Saved to: .specs-fire/intents/{intent-id}/brief.md

      ---

      Ready to break this into work items? [Y/n]
    </output>
    <check if="response == y">
      <invoke_skill>work-item-decompose</invoke_skill>
    </check>
  </step>
</flow>

<output_artifacts>

  | Artifact | Location | Template |
  |----------|----------|----------|
  | Intent Brief | `.specs-fire/intents/{id}/brief.md` | `./templates/brief.md.hbs` |
</output_artifacts>

<success_criteria>
  <criterion>User intent fully understood through dialogue</criterion>
  <criterion>Goal, users, problem clearly captured</criterion>
  <criterion>Success criteria defined</criterion>
  <criterion>Constraints identified</criterion>
  <criterion>Intent brief saved to correct location</criterion>
  <criterion>State.yaml updated with new intent</criterion>
</success_criteria>
COMPATIBILITY & REFERENCES

Platform guidance

Imported from the listed AgentSkill.sh category. Complete supplied package files and notices are retained. Required external tools and credentials are not installed. No bundled code was executed during import. Individual agent task behavior has not been tested.