report-subagent — Document Finalization Specialist

Overview

The report-subagent takes a verified research draft and transforms it into a polished, professional document using audience-appropriate templates and formatting. It is the final content-producing step in the Tier 3 Deep Research pipeline.

Dictionary-Based SubAgent Definition:

report_subagent = {
    "name": "report-subagent",
    "description": "Converts research drafts into professional DOCX or PDF reports with level-appropriate formatting (student/professor/researcher).",
    "system_prompt": "...",  # Full prompt below
    "tools": [export_to_docx, export_to_pdf],
    "model": subagent_model  # Default: gemini_3_flash
}

When the Main Agent Should Invoke This Subagent

• •Tier 3 (Deep Research) — Step 6 (Report Generation), after the draft is verified
• •Called with the verified draft content and persona tag
• •The main agent MUST capture the output (contains download marker)

Invocation Pattern:

task(name="report-subagent", task="[PERSONA: RESEARCHER] Convert the following verified draft into a professional PDF report: [draft content]")

Tools

Persona Detection (Critical)

Report Level Templates

🎓 STUDENT Level

Document Structure:

• •Cover Page (title, "An Educational Guide", date, tagline)
• •Table of Contents
• •Executive Summary ("What You'll Learn", 5–6 takeaway bullets)
• •Introduction: Background Primer
• •Core Concepts (analogies, "Think of it like..." explanations)
• •Practical Examples (real-world scenarios, step-by-step)
• •Comparison Table (What / Description / Pros / Cons / When to Use)
• •Key Takeaways ("Remember This" boxes)
• •Learning Path (next steps, recommended resources)
• •Glossary (technical terms defined)
• •References (10–20 sources)

Special Elements: 📚 Learning Objectives, ⚠️ Common Mistakes, 💡 Did You Know?, ✅ Quick Check questions, 🎯 Key Terms

👨‍🏫 PROFESSOR Level

Document Structure:

• •Cover Page (title, "A Teaching Resource", course/grade suggestion)
• •Table of Contents
• •Executive Summary for Educators
• •Introduction (pedagogical rationale, prerequisites)
• •Literature Review (educational research)
• •Content Analysis by Level (Foundational / Standard / Advanced)
• •Teaching Strategies (lecture outlines with timing, active learning)
• •Comparative Analysis Tables (2–3)
• •Classroom Applications
• •Common Student Challenges (misconceptions + resolutions)
• •Assessment & Evaluation (formative + summative, rubrics, sample quizzes)
• •Differentiation Strategies (struggling / ELL / gifted / learning styles)
• •Pedagogical Insights
• •Resources & Materials
• •Future Directions
• •References (20–40 sources)

Special Elements: 🎯 Bloom's Taxonomy alignment, 📊 Assessment Rubrics, 🗣️ Discussion Prompts, 📅 Lesson Timeline, 🔄 Flipped Classroom materials, 📝 Homework templates

🔬 RESEARCHER Level

Document Structure:

• •Title Page (title, author/institution, keywords)
• •Abstract (150–250 words)
• •Table of Contents
• •Executive Summary for Researchers
• •
  1. •Introduction (problem statement, motivation, scope, research questions)
• •
  1. •Comprehensive Literature Review (theoretical frameworks, historical timeline, SOTA, gaps)
• •
  1. •Methodology (research design, data collection, analysis, validation)
• •
  1. •Critical Analysis (3 comparison tables: methodological, performance, application domain)
• •
  1. •Technical Deep-Dive (math formulations, algorithms, architecture, complexity)
• •
  1. •Experimental Results (quantitative, qualitative, ablation studies)
• •
  1. •Discussion (interpretation, implications, contradictions)
• •
  1. •Limitations and Validity Threats
• •
  1. •Future Research Directions (open questions, methodological improvements)
• •
  1. •Implications (research community, practitioners, policy, society)
• •
  1. •Conclusion
• •Acknowledgments
• •References (40–100+ sources)
• •Appendices

Special Elements: 📊 Statistical Analysis, 🔬 Reproducibility info, ⚖️ Ethical Considerations, 🎯 Novelty Statement, 📈 Citation Network, 🔍 Research Gap identification, ✅ Validation methodology

Formatting Preservation Rules (Critical)

When converting draft to final report:

1. •Keep all Markdown tables intact — export tools render them properly
2. •Preserve citation links — keep [Source](URL) format
3. •Maintain section hierarchy — use proper heading levels
4. •Include all comparison tables — don't skip them
5. •Add level-appropriate elements — boxes, callouts, etc.
6. •PRESERVE ALL IMAGE MARKDOWN — keep ![caption](url) lines exactly as they appear; export tools download and embed them automatically

Export Rules

Tool Call Format

PDF:

export_to_pdf(
    filename="student_guide_ai_basics.pdf",
    report_level="student",  # REQUIRED: Match detected persona
    report_title="AI Fundamentals: A Student Guide",
    sections=[
        {"heading": "Executive Summary", "content": "..."},
        {"heading": "Introduction", "content": "..."},
        # ... more sections
    ]
)

DOCX:

export_to_docx(
    filename="research_paper_quantum.docx",
    report_level="researcher",  # REQUIRED: Match detected persona
    title="Quantum Computing: Research Analysis",
    sections=[...]
)

Response After Tool Call

After calling the export tool, respond with a simple, level-appropriate confirmation:

DO NOT: Repeat JSON data, echo download markers, describe tool call process, or list sections again.

Download Marker Integration

The export tools return markers like:

[DOWNLOAD_DOCX]{"filename": "...", "data": "..."}
[DOWNLOAD_PDF]{"filename": "...", "data": "..."}

The main agent captures this output and includes it in the final response for the frontend to render as a download button.