interview

name	interview
description	Run a Socratic Interview loop to clarify requirements, or manage interview sessions.

Run OmA interview command.

Input: $ARGUMENTS

Protocol:

1. Subcommand Handling (Optional):

   • If the first argument is a subcommand, execute the corresponding action:
     • status: Read .omg/state/interviews/active.json, then load the referenced .omg/state/interviews/[slug]/context.json. Display current Clarity Score, Confirmed Facts, and Ready-to-Run Prompt based ONLY on that file content.
     • help: Display instructions on how the interview works.
     • resume: Read .omg/state/interviews/active.json, then load the referenced session context. If the pointer or context file is missing, tell the user no active session exists and they must run /oma:intent [depth] first.
     • done: Finalize interview, generate PRD, and transition to ready-to-execute.
     • cancel: Invoke OmA cancel flow specifically for the interview session to stop safely and save checkpoint artifacts.
     • restart: Clear .omg/state/interviews/active.json and start a new session folder.

2. Default Interview Loop (If no subcommand):

   • Language Alignment:
     • Detect the language used by the user in the current session.
     • All outputs (Confirmed Facts, Socratic Questions, and status updates) MUST be written in the user's detected language.
   • Depth: Default is low if no depth keyword is provided; otherwise, use the provided depth keyword (e.g., medium, high).
   • Initialize or Resume (Strict File-Based Logic):
     • Check First: Use read_file to verify the existence of .omg/state/interviews/active.json.
     • If the file is missing: Auto-initialize a new session folder, create .omg/state/interviews/active.json pointing to it, and seed the initial .omg/state/interviews/[slug]/context.json with the current user request.
     • If resuming: Resolve the active session folder from .omg/state/interviews/active.json, then resolve the session context. This context is the ONLY source of truth.
   • Socratic Dialogue:
     • Invoke the interview agent.
     • Identify the highest-risk ambiguity or technical gap.
     • Ask ONE targeted Socratic question in the user's language.
   • Convergence & Scoring:
     • Update clarity_score, facts, and questions_asked in .omg/state/interviews/[slug]/context.json after every turn.
     • Scoring Criteria (0-100):
       • Per question (turn), award +5 to exactly ONE of the following criteria:
         1. Core objective defined
         2. Target audience/context clear
         3. Tech stack/constraints identified
         4. Edge cases considered
         5. Acceptance criteria set
     • Termination Criteria:
       • Always continue until Clarity Score >= 100.
     • Threshold Gates (for [READY-TO-RUN PROMPT]):
       • low: 10, medium: 20, high: 30, xhigh: 40, ultra: 50
   • Finalization:
     • Automatically trigger on Clarity Score >= 100.
     • Manual Trigger by done or $intent-done:
       • Generate .omg/state/interviews/[slug]/prd.md based on .omg/templates/prd-template.md.
       • Transition the workflow to ready_to_execute.
       • Instruct the user to run /oma:team-assemble --intent="[Confirmed Objective]" or /oma:team-exec --intent="[Confirmed Objective]" (the intent description SHOULD be in English for system consistency).
     • Triggered by cancel:
       • Invoke OmA cancel flow specifically for the interview session to stop safely and save checkpoint artifacts (status: stopped, no transition).

Response:

• Keep it concise and professional.
• Use the user's detected language for all content.
• Include [PROGRESS], [CONFIRMED FACTS], [SOCRATIC QUESTION], and the [READY-TO-RUN PROMPT].
• Note: If Clarity Score < Threshold for the selected depth, set [READY-TO-RUN PROMPT] to not yet. Unlock at [Threshold Score] points (translated if appropriate).
• If Clarity Score >= Threshold, the [READY-TO-RUN PROMPT] MUST include a command (e.g., /oma:team-assemble --intent="...") reflecting current facts and a specific question in the user's language asking if they want to implement this now.