interview
Scannednpx machina-cli add skill Q00/ouroboros/interview --openclaw/ouroboros:interview
Socratic interview to crystallize vague requirements into clear specifications.
Usage
ooo interview [topic]
/ouroboros:interview [topic]
Trigger keywords: "interview me", "clarify requirements"
Instructions
When the user invokes this skill, choose the execution path:
Path A: MCP Mode (Preferred)
If the ouroboros_interview MCP tool is available, use it for persistent, structured interviews:
-
Start a new interview:
Tool: ouroboros_interview Arguments: initial_context: <user's topic or idea>The tool returns a session ID and the first question.
-
Present the question using AskUserQuestion: After receiving a question from the tool, present it via
AskUserQuestionwith contextually relevant suggested answers:{ "questions": [{ "question": "<question from MCP tool>", "header": "Q<N>", "options": [ {"label": "<option 1>", "description": "<brief explanation>"}, {"label": "<option 2>", "description": "<brief explanation>"} ], "multiSelect": false }] }Generating options β analyze the question and suggest 2-3 likely answers:
- Binary questions (greenfield/brownfield, yes/no): use the natural choices
- Technology choices: suggest common options for the context
- Open-ended questions: suggest representative answer categories
- The user can always type a custom response via "Other"
-
Relay the answer back:
Tool: ouroboros_interview Arguments: session_id: <session ID from step 1> answer: <user's selected option or custom text>The tool records the answer, generates the next question, and returns it.
-
Repeat steps 2-3 until the user says "done" or requirements are clear.
-
After completion, suggest
ooo seedto generate the Seed specification.
Advantages of MCP mode: State persists to disk (survives session restarts), ambiguity scoring, direct integration with ooo seed via session ID, structured input with AskUserQuestion.
Path B: Plugin Fallback (No MCP Server)
If the MCP tool is NOT available, fall back to agent-based interview:
- Read
agents/socratic-interviewer.mdand adopt that role - Ask clarifying questions based on the user's topic
- Present each question using AskUserQuestion with contextually relevant suggested answers (same format as Path A step 2)
- Use Read, Glob, Grep, WebFetch to explore context if needed
- Continue until the user says "done"
- Interview results live in conversation context (not persisted)
Interviewer Behavior (Both Modes)
The interviewer is ONLY a questioner:
- Always ends responses with a question
- Targets the biggest source of ambiguity
- NEVER writes code, edits files, or runs commands
Example Session
User: ooo interview Build a REST API
Q1: What domain will this REST API serve?
User: It's for task management
Q2: What operations should tasks support?
User: Create, read, update, delete
Q3: Will tasks have relationships (e.g., subtasks, tags)?
User: Yes, tags for organizing
User: ooo seed [Generate seed from interview]
Next Steps
After interview completion, use ooo seed to generate the Seed specification.
Overview
This skill guides a structured Socratic dialogue to turn vague ideas into clear specifications. It supports two execution paths: a persistent MCP-based interview when available, or a plugin fallback. The interviewer always ends each response with a clarifying question to resolve ambiguity.
How This Skill Works
On invocation, the tool selects Path A (MCP mode) if the ouroboros_interview tool is available, starting a new session and delivering questions via AskUserQuestion payloads. It records answers, generates follow-up questions, and persists state; if MCP is unavailable, it uses a non-persistent, question-only interview. When the interview completes, it suggests running ooo seed to generate the Seed specification.
When to Use It
- When you have a topic or idea but vague requirements
- When stakeholders disagree or provide conflicting details
- When you want a repeatable, structured clarification process
- When you need a documented sequence of clarifying questions
- When youβre preparing input for a Seed specification
Quick Start
- Step 1: Run 'ooo interview <topic>' to start
- Step 2: Answer questions or choose from suggested options; the flow may persist via MCP
- Step 3: Once complete, run 'ooo seed' to generate the Seed specification
Best Practices
- Start with a plainly stated topic to anchor the interview
- Prefer MCP mode for persistent sessions and structured input
- End each user response with a clarifying question to reduce ambiguity
- Use the provided suggested options to guide binary or multiple-choice answers
- Document the session ID or seed output for traceability
Example Use Cases
- Clarifying the scope of a new task management REST API
- Defining requirements for a data visualization dashboard
- Elaborating user roles and permissions for an internal tool
- Specifying minimum viable features for a mobile app
- Iteratively refining non-functional requirements (performance, reliability)