Final Response

Usage

report = final_response_skill(
    title="Reconnaissance Analysis Report",
    summary="Analysis of drone footage reveals significant enemy armor activity...",
    findings=[
        "3 tanks detected moving eastward in sector Alpha",
        "Convoy formation suggests defensive posture",
        {"text": "Infantry support observed", "confidence": "high", "source": "video"}
    ],
    video_evidence=[
        {
            "video_id": "m-abc123",
            "segment_id": 2,
            "time_range": "00:00:30 - 00:01:00",
            "description": "Tank formation in open terrain",
            "objects": ["tank", "tank", "truck"]
        }
    ],
    pdf_references=[
        {
            "source": "field_manual.pdf",
            "text": "T-72 tanks have operational range of 500km...",
            "relevance_score": 0.85
        }
    ],
    tactical_assessment="The observed enemy movement pattern indicates preparation for defensive operations...",
    recommendations=[
        "Monitor sector Alpha for continued activity",
        "Prepare counter-measures for potential armor engagement",
        "Request aerial reconnaissance for broader coverage"
    ],
    methodology="Semantic search and object detection across 2 videos, cross-referenced with tactical doctrine"
)

Parameters

Required

• title: Report title (e.g., "Tank Activity Analysis Report")
• summary: Executive summary (2-3 sentences)
• findings: List of key findings (strings or dicts with text/confidence/source)

Optional

• video_evidence: List of video evidence items (dicts with video_id, segment_id, time_range, description, objects)
• pdf_references: List of PDF citations (dicts with source, text, relevance_score)
• tactical_assessment: Overall tactical assessment paragraph
• recommendations: List of recommended actions
• methodology: Description of analysis approach

Finding Format Options

Findings can be simple strings or structured dicts:

# Simple string
"3 tanks detected in sector Alpha"

# Structured dict
{
    "text": "3 tanks detected in sector Alpha",
    "confidence": "high",
    "source": "video segment 2"
}

Video Evidence Format

{
    "video_id": "m-abc123xyz",
    "segment_id": 2,
    "time_range": "00:00:30 - 00:01:00",
    "description": "Tank formation moving through open terrain",
    "objects": ["tank", "tank", "truck"]
}

PDF Reference Format

{
    "source": "tank_manual.pdf",
    "text": "Relevant excerpt from the document...",
    "relevance_score": 0.85
}

Result Structure

Returns:

• status: "success" or "error"
• result:
  • report: Formatted markdown report
  • sections: Metadata about included sections
• message: Summary of report contents

Example Workflow

# 1. Gather evidence using other skills
contexts = videodb_query_skill("get_contexts")
tanks = videodb_query_skill("object_search", object_type="tank")
specs = pdf_rag_skill("tank capabilities")
situation = wargame_query_skill("tactical_situation")

# 2. Synthesize findings
findings = []
video_evidence = []
pdf_references = []

# Process tank results
if tanks["status"] == "success":
    findings.append(tanks["result"]["summary_text"])
    for entry in tanks["result"]["timeline"]:
        video_evidence.append({
            "video_id": entry["video_id"],
            "segment_id": entry["segment_id"],
            "time_range": f"{entry['start_time_hms']} - {entry['end_time_hms']}",
            "description": entry.get("description", ""),
            "objects": [f"tank x{entry['object_count']}"]
        })

# Process PDF results
if specs["status"] == "success":
    for r in specs["result"]["results"][:3]:
        pdf_references.append({
            "source": r["source"],
            "text": r["text"],
            "relevance_score": r["relevance_score"]
        })

# 3. Generate final report
report = final_response_skill(
    title="Tank Activity Analysis",
    summary="Comprehensive analysis of detected tank activity...",
    findings=findings,
    video_evidence=video_evidence,
    pdf_references=pdf_references,
    tactical_assessment="Based on the evidence...",
    recommendations=["Continue monitoring", "Alert command"]
)

# 4. Return the formatted report
final_answer(report["result"]["report"])

Best Practices

1. Always use for final answers - Maintains consistency
2. Include methodology - Explain how you analyzed
3. Cite all sources - Video segments, PDFs, map data
4. Be specific - Include segment IDs, time ranges
5. Actionable recommendations - Provide clear next steps
6. Causal relationships - Connect evidence to conclusions