Grill Me

{
  "schema_version": "2.0",
  "meta": {
    "generated_at": "2026-03-27T08:24:21.532Z",
    "slug": "mattpocock-grill-me",
    "source_url": "https://github.com/mattpocock/skills/tree/main/grill-me/",
    "source_ref": "main",
    "model": "claude",
    "analysis_version": "3.0.0",
    "source_type": "community",
    "content_hash": "2630167f799732d4b3bb873b87729215359b79112f936594bf8286cbace21030",
    "tree_hash": "ddafdc02b5ad24d111480e6bad9058cbab110321c2e52c4e59817339b8b09031"
  },
  "skill": {
    "name": "grill-me",
    "description": "Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions \"grill me\".",
    "summary": "Conducts rigorous design interviews to expose hidden flaws and dependencies in your plans before implementation.",
    "icon": "📦",
    "version": "1.0.0",
    "author": "mattpocock",
    "license": "MIT",
    "category": "productivity",
    "tags": [
      "design review",
      "planning",
      "interview",
      "decision-making",
      "validation"
    ],
    "supported_tools": [
      "claude",
      "codex",
      "claude-code"
    ],
    "risk_factors": []
  },
  "security_audit": {
    "risk_level": "safe",
    "is_blocked": false,
    "safe_to_publish": true,
    "summary": "Static analyzer flagged false positives for 'weak cryptographic algorithm' on lines 3 and 6 of SKILL.md. These lines contain plain English instructional text about interviewing users, with no cryptographic code present. The skill is purely conversational with no code execution, network access, or file operations. Safe for publication.",
    "risk_factor_evidence": [],
    "critical_findings": [],
    "high_findings": [],
    "medium_findings": [],
    "low_findings": [],
    "dangerous_patterns": [],
    "files_scanned": 1,
    "total_lines": 11,
    "audit_model": "claude",
    "audited_at": "2026-03-27T08:24:21.532Z"
  },
  "content": {
    "user_title": "Stress-test your plans with relentless interview questions",
    "value_statement": "Uncritical planning leads to overlooked flaws and failed implementations. This skill rigorously interviews your designs, exposing hidden dependencies and weak points before they become costly mistakes.",
    "seo_keywords": [
      "design review",
      "plan validation",
      "Claude interview skill",
      "Codex design helper",
      "Claude Code planning",
      "technical interview",
      "architecture review",
      "decision tree analysis",
      "design thinking",
      "requirements gathering"
    ],
    "actual_capabilities": [
      "Conducts systematic one-question-at-a-time interviews about your plans",
      "Explores your codebase to answer factual questions when needed",
      "Identifies hidden dependencies between design decisions",
      "Provides recommended answers for each decision point",
      "Walks through decision trees branch by branch until shared understanding is reached"
    ],
    "limitations": [
      "Cannot execute code or run tests to validate assumptions",
      "Relies on user providing accurate context and honest answers",
      "Interview depth depends on user engagement and time investment",
      "Cannot access external systems or documentation beyond provided codebase"
    ],
    "use_cases": [
      {
        "title": "Architecture Review Before Implementation",
        "description": "Validate your system architecture by answering probing questions about component interactions, scalability concerns, and failure modes before writing any code.",
        "target_user": "Senior developers and architects planning new systems"
      },
      {
        "title": "Design Document Stress-Testing",
        "description": "Expose gaps in your design documentation by defending each decision against targeted questions that reveal unstated assumptions and missing edge cases.",
        "target_user": "Technical leads preparing RFCs or design proposals"
      },
      {
        "title": "Learning Through Guided Discovery",
        "description": "Deepen your understanding of a domain by being interviewed through its key concepts and decision points, with recommendations provided at each step.",
        "target_user": "Developers exploring unfamiliar technical domains"
      }
    ],
    "prompt_templates": [
      {
        "title": "Basic Plan Review",
        "prompt": "I'm planning to build a feature that allows users to export their data as CSV files. The plan is to add a button that triggers a backend endpoint which queries the database and streams the response. Grill me on this plan."
      },
      {
        "title": "Architecture Decision Interview",
        "prompt": "I'm deciding between using Redis or in-memory caching for our session store. We expect 10k concurrent users and need sub-100ms response times. Ask me questions to help me make the right choice, one question at a time."
      },
      {
        "title": "Codebase Exploration Integration",
        "prompt": "I want to add authentication to our existing API. Before I start coding, explore our codebase to understand the current structure, then grill me on the implementation approach. Start by finding our existing middleware patterns."
      },
      {
        "title": "Complex System Design Validation",
        "prompt": "I'm designing a multi-tenant SaaS platform with isolated databases per tenant, shared application servers, and a unified monitoring system. Grill me on every aspect of this architecture including data isolation, scaling strategies, backup procedures, and failure recovery. Provide your recommended answer for each question before moving to the next."
      }
    ],
    "output_examples": [
      {
        "input": "I want to add real-time notifications to our app using WebSockets.",
        "output": [
          "First question: What is your expected concurrent connection count at peak load, and have you validated that your current infrastructure can handle persistent WebSocket connections at that scale?",
          "Recommended answer: For most applications under 5k concurrent users, a single server with WebSocket support may suffice. Above that, you will need a message broker like Redis Pub/Sub to distribute messages across multiple application servers. Consider whether your hosting environment supports long-lived connections."
        ]
      },
      {
        "input": "We are migrating from monolith to microservices. I think we should split by domain.",
        "output": [
          "First question: What specific pain points in your current monolith are you trying to solve with microservices, and have you measured their impact quantitatively?",
          "Recommended answer: Common drivers include independent deployment needs, team scaling issues, or specific components requiring different scaling characteristics. If you cannot point to concrete problems that microservices solve, the migration complexity may outweigh the benefits. Consider starting with modular monolith patterns first."
        ]
      }
    ],
    "best_practices": [
      "Answer each question honestly and completely before moving to the next one",
      "Provide context about your constraints and requirements upfront for more relevant questions",
      "Use the codebase exploration feature when questions involve existing code patterns"
    ],
    "anti_patterns": [
      "Skipping questions or providing vague answers that hide real concerns",
      "Rushing through the interview without reflecting on each decision point",
      "Ignoring the recommended answers without considering the trade-offs they highlight"
    ],
    "faq": [
      {
        "question": "How long does a typical grill session take?",
        "answer": "Sessions vary based on plan complexity. Simple feature plans may take 10-15 questions, while architecture reviews can span 30+ questions over multiple conversations. You can pause and resume at any time."
      },
      {
        "question": "Can this skill access my private codebase?",
        "answer": "Yes, when you ask questions about your code, the skill can explore your repository to understand existing patterns and provide context-aware recommendations."
      },
      {
        "question": "What if I disagree with a recommended answer?",
        "answer": "Recommended answers are suggestions based on common patterns. You can explain your reasoning for a different approach, and the interview will adapt to explore your alternative."
      },
      {
        "question": "Is this skill suitable for beginner developers?",
        "answer": "Yes. The interview format helps beginners learn to think through designs systematically. Recommended answers provide educational context at each step."
      },
      {
        "question": "Can I use this for non-technical planning?",
        "answer": "While designed for technical decisions, the interview approach works for any structured planning scenario. The codebase exploration feature is specific to software projects."
      },
      {
        "question": "How do I get the most value from a grill session?",
        "answer": "Be honest about uncertainties, explore recommended answers even when they challenge your assumptions, and use the one-question-at-a-time format to fully think through each decision before moving forward."
      }
    ]
  },
  "file_structure": [
    {
      "name": "SKILL.md",
      "type": "file",
      "path": "SKILL.md",
      "lines": 11
    }
  ]
}