{
  "task": {
    "title": "AI Code Assistants Impact on Developer Productivity: Evidence and Best Practices",
    "objective": "Evaluate the documented productivity impact of AI code assistants (GitHub Copilot, Cursor, Codeium, etc.) and identify optimal adoption strategies for engineering organizations",
    "type": "comparative"
  },
  "context": {
    "background": "AI code assistants have seen rapid adoption since GitHub Copilot's 2022 launch. Organizations report productivity gains but measurement methodologies vary. Security and code quality concerns persist alongside adoption.",
    "audience": "technical",
    "use_case": "Engineering leadership evaluating AI assistant rollout for 500-person engineering org",
    "prior_knowledge": [
      "GitHub Copilot claims 55% faster task completion",
      "Major concerns include security leaks and code quality",
      "Adoption varies significantly by task type"
    ]
  },
  "questions": {
    "primary": "What is the evidence-based productivity impact of AI code assistants, and what factors predict successful organizational adoption?",
    "secondary": [
      "What metrics are used to measure productivity impact in published studies?",
      "How does impact vary by programming language and task type?",
      "What are documented security risks and mitigation strategies?",
      "How do different tools (Copilot, Cursor, Codeium, etc.) compare?",
      "What organizational change management practices correlate with successful adoption?",
      "What is the impact on code quality and technical debt?"
    ],
    "hypotheses": [
      "Productivity gains are highest for boilerplate/repetitive code",
      "Senior developers benefit more than juniors",
      "Impact is lower for complex algorithmic tasks"
    ],
    "exclusions": [
      "General LLM capabilities (ChatGPT for coding questions)",
      "Code review tools (unless AI-powered)",
      "Testing automation tools",
      "DevOps/CI-CD tools"
    ]
  },
  "constraints": {
    "timeframe": {
      "start": "2022-01-01",
      "end": "present",
      "focus_period": "2024"
    },
    "geography": {
      "scope": "global",
      "regions": [],
      "exclude_regions": []
    },
    "sources": {
      "required_types": ["peer_reviewed", "industry_reports", "official_docs", "case_studies"],
      "preferred_domains": ["github.blog", "arxiv.org", "acm.org", "ieee.org"],
      "excluded_domains": ["medium.com"],
      "min_quality": "B",
      "language": ["en"]
    },
    "data_requirements": {
      "quantitative": true,
      "qualitative": true,
      "specific_metrics": [
        "task completion time reduction",
        "lines of code accepted vs suggested",
        "developer satisfaction scores",
        "code quality metrics",
        "security incident rates"
      ]
    }
  },
  "output": {
    "format": "comprehensive_report",
    "length": {
      "min_words": 4000,
      "max_words": 10000,
      "executive_summary_words": 400
    },
    "structure": {
      "include_executive_summary": true,
      "include_methodology": true,
      "include_visualizations": true,
      "include_raw_data": false,
      "include_bibliography": true,
      "include_appendices": true,
      "generate_website": false
    },
    "citation_style": "IEEE",
    "tone": "technical"
  },
  "keywords": [
    "GitHub Copilot",
    "AI code assistant",
    "developer productivity",
    "code completion",
    "Cursor AI",
    "Codeium",
    "coding AI",
    "software development AI"
  ],
  "special_instructions": [
    "Compare at least 3 major tools with feature matrices",
    "Separate internal studies (vendor research) from independent research",
    "Include adoption best practices section with actionable recommendations",
    "Note methodological limitations of cited productivity studies",
    "Address security and IP concerns with specific mitigation strategies"
  ]
}