3 KiB
3 KiB
Yanstaller Interview Workflow
Mode: Non-interactive (JSON output) Model: gpt-5-mini Durée: 2-3 minutes
Input Format
Prompt: interview:<projet_type>:<domaine>:<equipe_size>:<experience>
Examples:
interview:new:web:solo:beginnerinterview:existing:backend:medium:expert
Interview Questions
Phase 1: Project Context
- Type de projet? (new/existing/migration)
- Domaine? (web/backend/data/mobile/devops/other)
- Taille équipe? (solo/small/medium/large)
- Expérience AI? (beginner/intermediate/expert)
Phase 2: Environment
- Connectivity? (online/offline/intermittent)
- GPU disponible? (yes/no/unknown)
Phase 3: Goals
- Objectifs? (agents/workflows/tests/analysis/voice)
- Méthodologie? (agile/tdd/merise-agile/hybrid)
- Fréquence? (daily/weekly/occasional)
- Qualité? (mvp/balanced/production/critical)
Output Format (JSON)
{
"interview_completed": true,
"responses": {
"projectType": "new",
"domain": "web",
"teamSize": "solo",
"experience": "expert",
"connectivity": "online",
"gpuAvailable": "yes",
"objectives": ["agents", "workflows", "voice"],
"methodology": "merise-agile",
"frequency": "daily",
"qualityLevel": "balanced"
},
"recommendations": {
"platforms": [
{
"name": "copilot",
"reason": "detected + large community",
"priority": 1
},
{
"name": "codex",
"reason": "detected + workflow-focused",
"priority": 2
}
],
"turboWhisper": {
"recommended": true,
"mode": "docker",
"reason": "GPU available + daily usage"
},
"agents": {
"essential": ["byan", "analyst", "architect"],
"optional": ["dev", "pm", "sm"],
"reason": "Solo expert with full-stack needs"
},
"methodology": {
"name": "merise-agile",
"modules": ["bmm", "bmb", "tea"],
"reason": "Native BYAN approach"
}
},
"complexity_score": 15,
"recommended_model": "gpt-5-mini"
}
Logic
Platform Recommendations
- Priority 1: All detected platforms
- Priority 2: If no detection, recommend Copilot CLI (most popular)
Turbo Whisper Recommendations
- Yes GPU + daily: docker mode
- No GPU + daily: local mode (with warning: slower)
- Not daily: skip (can install later)
Agent Recommendations
- Beginner: 1-2 agents (byan only)
- Intermediate: 3-5 agents (byan + analyst + pm)
- Expert: All agents (full toolbox)
Methodology Mapping
- TDD → Tea module priority
- Merise Agile → BMM full workflow
- Agile/Hybrid → Quick-flow prioritized
Usage in NPX
const result = execSync(`copilot --agent=bmad-agent-yanstaller --prompt "interview" --model gpt-5-mini`);
const json = JSON.parse(extractJSON(result));
// Apply recommendations
const platform = json.recommendations.platforms[0].name;
const turboMode = json.recommendations.turboWhisper.mode;