Output Formats¶

PlanOpticon produces a wide range of output formats from video analysis, document ingestion, batch processing, knowledge graph export, and agent skills. This page is the comprehensive reference for every format the tool can emit.

Transcripts¶

Video analysis always produces transcripts in three formats, stored in the transcript/ subdirectory of the output folder.

Transcript JSON structure¶

{
  "segments": [
    {
      "start": 0.0,
      "end": 4.5,
      "text": "Welcome to the sprint review.",
      "speaker": "Alice"
    }
  ],
  "text": "Welcome to the sprint review. ...",
  "language": "en"
}

When the --speakers flag is provided (e.g., --speakers "Alice,Bob,Carol"), speaker diarization hints are passed to the transcription provider and speaker labels appear in the JSON segments.

Reports¶

Analysis reports are generated from the combined transcript, diagrams, key points, action items, and knowledge graph. They live in the results/ subdirectory.

Diagrams¶

Each visual element detected during frame analysis produces up to five output files in the diagrams/ subdirectory. The index N is zero-based.

Frames that score as medium confidence are saved as captioned screenshots in the captures/ subdirectory instead, with a capture_N.jpg and capture_N.json pair.

Structured Data¶

Core analysis artifacts are stored as JSON files in the results/ subdirectory.

Knowledge graph JSON structure¶

{
  "entities": [
    {
      "name": "Kubernetes",
      "type": "technology",
      "descriptions": ["Container orchestration platform discussed in architecture review"],
      "occurrences": [
        {"source": "video:recording.mp4", "timestamp": "00:05:23"}
      ]
    }
  ],
  "relationships": [
    {
      "source": "Kubernetes",
      "target": "Docker",
      "type": "DEPENDS_ON",
      "descriptions": ["Kubernetes uses Docker as container runtime"]
    }
  ]
}

Charts¶

When chart data is extracted from diagrams (bar charts, line charts, pie charts, scatter plots), PlanOpticon reproduces them as standalone image files.

Reproduced charts are also embedded inline in the HTML and PDF reports.

Knowledge Graph Exports¶

Beyond the default knowledge_graph.db and knowledge_graph.json produced during analysis, PlanOpticon supports exporting knowledge graphs to several additional formats via the planopticon export and planopticon kg convert commands.

Converting between formats¶

# SQLite to JSON
planopticon kg convert results/knowledge_graph.db output.json

# JSON to SQLite
planopticon kg convert knowledge_graph.json knowledge_graph.db

# Sync both directions (updates the stale file)
planopticon kg sync results/knowledge_graph.db
planopticon kg sync knowledge_graph.db knowledge_graph.json --direction db-to-json

PlanOpticonExchange Format¶

The PlanOpticonExchange format (.json) is a canonical interchange payload designed for sharing knowledge graphs between PlanOpticon instances, teams, or external systems.

planopticon export exchange knowledge_graph.db
planopticon export exchange kg.db -o exchange.json --name "My Project"

The exchange payload includes:

• Schema version for forward compatibility
• Project metadata (name, description)
• Full entity and relationship data with provenance
• Source tracking for multi-source graphs
• Merge support -- exchange files can be merged together, deduplicating entities by name

Exchange JSON structure¶

{
  "schema_version": "1.0",
  "project": {
    "name": "Sprint Reviews Q4",
    "description": "Knowledge extracted from Q4 sprint review recordings"
  },
  "entities": [...],
  "relationships": [...],
  "sources": [...]
}

Document Exports¶

PlanOpticon can generate structured Markdown documents from any knowledge graph, with no API key required. These are pure template-based outputs derived from the graph data.

Markdown document types¶

There are seven document types plus a CSV export, all generated via planopticon export markdown:

# Generate all document types
planopticon export markdown knowledge_graph.db

# Generate specific types
planopticon export markdown kg.db -o ./docs --type summary --type glossary

# Generate meeting notes and CSV
planopticon export markdown kg.db --type meeting-notes --type csv

Obsidian vault export¶

Exports the knowledge graph as an Obsidian-compatible vault with YAML frontmatter, [[wiki-links]] between entities, and proper folder structure.

planopticon export obsidian knowledge_graph.db -o ./my-vault

The vault includes:

• One note per entity with frontmatter (type, aliases, tags)
• Wiki-links between related entities
• A _Index.md file for navigation
• Compatible with Obsidian graph view

Notion markdown export¶

Exports as Notion-compatible Markdown with a CSV database file for import into Notion databases.

planopticon export notion knowledge_graph.db -o ./notion-export

GitHub wiki export¶

Generates a complete GitHub wiki with a sidebar, home page, and per-entity pages. Can be pushed directly to a GitHub wiki repository.

# Generate wiki pages
planopticon wiki generate knowledge_graph.db -o ./wiki

# Push to GitHub
planopticon wiki push ./wiki ConflictHQ/PlanOpticon -m "Update wiki from KG"

Batch Outputs¶

Batch processing produces additional files at the batch root directory, alongside per-video output folders.

Self-Contained HTML Viewer¶

PlanOpticon ships with a self-contained interactive knowledge graph viewer at knowledge-base/viewer.html in the repository. This file:

• Uses D3.js (bundled inline, no CDN dependency)
• Renders an interactive force-directed graph visualization
• Supports node filtering by entity type
• Shows entity details and relationships on click
• Can load any knowledge_graph.json file
• Works offline with no server required -- just open in a browser
• Covers approximately 80% of graph exploration needs with zero infrastructure

Output Directory Structure¶

A complete single-video analysis produces the following directory tree:

output/
├── manifest.json                      # Run manifest (source of truth)
├── transcript/
│   ├── transcript.json                # Full transcript with segments
│   ├── transcript.txt                 # Plain text
│   └── transcript.srt                 # Subtitles
├── frames/
│   ├── frame_0000.jpg                 # Extracted video frames
│   ├── frame_0001.jpg
│   └── ...
├── diagrams/
│   ├── diagram_0.jpg                  # Original frame
│   ├── diagram_0.mermaid             # Mermaid source
│   ├── diagram_0.svg                 # Vector rendering
│   ├── diagram_0.png                 # Raster rendering
│   ├── diagram_0.json                # Analysis data
│   ├── chart_0.svg                   # Reproduced chart (SVG)
│   ├── chart_0.png                   # Reproduced chart (PNG)
│   └── ...
├── captures/
│   ├── capture_0.jpg                 # Medium-confidence screenshots
│   ├── capture_0.json                # Caption and metadata
│   └── ...
└── results/
    ├── analysis.md                    # Markdown report
    ├── analysis.html                  # HTML report
    ├── analysis.pdf                   # PDF report (if planopticon[pdf] installed)
    ├── knowledge_graph.db             # Knowledge graph (SQLite, primary)
    ├── knowledge_graph.json           # Knowledge graph (JSON export)
    ├── key_points.json                # Extracted key points
    └── action_items.json              # Action items

Controlling Output Format¶

Use the --output-format flag with planopticon analyze to control how results are presented:

# Standard output (files + console summary)
planopticon analyze -i video.mp4 -o ./output

# JSON manifest to stdout (for scripting)
planopticon analyze -i video.mp4 -o ./output --output-format json