PlanOpticon

AI-powered video analysis, knowledge extraction, and planning.

PlanOpticon processes video recordings, documents, and 20+ online sources into structured knowledge graphs, then helps you plan with an AI agent and interactive companion. It auto-discovers models across 15+ AI providers, runs fully offline with Ollama, and produces rich multi-format output.

Features

• 15+ AI providers -- OpenAI, Anthropic, Gemini, Ollama, Azure, Together, Fireworks, Cerebras, xAI, Bedrock, Vertex, Mistral, Cohere, AI21, HuggingFace, Qianfan, and LiteLLM. Defaults to cheap models (Haiku, GPT-4o-mini, Gemini Flash).
• 20+ source connectors -- YouTube, web pages, GitHub, Reddit, HackerNews, RSS, podcasts, arXiv, S3, Google Workspace, Microsoft 365, Obsidian, Notion, Apple Notes, Zoom, Teams, Google Meet, and more.
• Planning agent -- 11 skills including project plans, PRDs, roadmaps, task breakdowns, and GitHub integration.
• Interactive companion -- Chat REPL with 15 slash commands, auto-discovery of workspace knowledge, and runtime provider/model switching.
• Knowledge graphs -- SQLite-backed (zero external deps), entity extraction with planning taxonomy (goals, requirements, risks, tasks, milestones), merge and dedup across sources.
• Smart video analysis -- Change-detection frame extraction, face filtering, diagram classification, action item detection, checkpoint/resume.
• Document ingestion -- PDF, Markdown, and plaintext pipelines feed the same knowledge graph.
• Export everywhere -- Markdown docs (7 types, no LLM required), Obsidian vaults, Notion markdown, GitHub wiki with push, PlanOpticonExchange JSON interchange, HTML/PDF reports, Mermaid diagrams.
• OAuth-first auth -- Unified OAuth manager for Google, Dropbox, Zoom, Notion, GitHub, and Microsoft with saved-token / PKCE / API-key fallback chain.
• Batch processing -- Process entire folders with merged knowledge graphs and cross-referencing.

Quick Start

# Install
pip install planopticon

# Analyze a video
planopticon analyze -i meeting.mp4 -o ./output

# Ingest a document
planopticon ingest -i spec.pdf -o ./output

# Fetch from a source
planopticon fetch youtube "https://youtube.com/watch?v=..." -o ./output

# Process a folder of videos
planopticon batch -i ./recordings -o ./output --title "Weekly Meetings"

# Query the knowledge graph
planopticon query
planopticon query "entities --type technology"

# See available AI models
planopticon list-models

Planning Agent

Run AI-powered planning skills against your knowledge base:

# Generate a project plan from extracted knowledge
planopticon agent "Create a project plan" --kb ./results

# Build a PRD
planopticon agent "Write a PRD for the authentication system" --kb ./results

# Break down tasks
planopticon agent "Break this into tasks and estimate effort" --kb ./results

11 skills: project_plan, prd, roadmap, task_breakdown, github_integration, requirements_chat, doc_generator, artifact_export, cli_adapter, notes_export, wiki_generator.

Interactive Companion

A chat REPL that auto-discovers knowledge graphs, videos, and docs in your workspace:

# Launch the companion
planopticon companion
# or
planopticon --chat

15 slash commands: /help, /status, /skills, /entities, /search, /neighbors, /export, /analyze, /ingest, /auth, /provider, /model, /run, /plan, /prd, /tasks.

Switch providers and models at runtime, explore your knowledge graph interactively, or chat with any configured LLM.

Source Connectors

Export & Documents

Generate documents from your knowledge graph without an LLM:

planopticon export summary -o ./docs
planopticon export meeting-notes -o ./docs
planopticon export glossary -o ./docs

7 document types: summary, meeting-notes, glossary, relationship-map, status-report, entity-index, csv.

Additional export targets: - Obsidian -- YAML frontmatter + wiki-links vault - Notion -- Compatible markdown - GitHub Wiki -- Generate and push directly - PlanOpticonExchange -- Canonical JSON interchange with merge/dedup

Local Run

PlanOpticon runs entirely offline with Ollama -- no API keys, no cloud, no cost.

13.2 hours of video content analyzed, knowledge-graphed, and summarized in ~25 hours of processing time, entirely on local hardware, for free.

18 meeting recordings processed on a single machine using llava (vision), qwen3:30b (chat), and whisper-large (transcription via Apple Silicon GPU):

# Fully local analysis -- no API keys needed, just Ollama running
planopticon analyze -i meeting.mp4 -o ./output \
  --provider ollama \
  --vision-model llava:latest \
  --chat-model qwen3:30b

Installation

From PyPI

pip install planopticon

# With all extras (PDF, cloud sources, GPU)
pip install planopticon[all]

From Source

git clone https://github.com/ConflictHQ/PlanOpticon.git
cd PlanOpticon
pip install -e ".[dev]"

Binary Download

Download standalone binaries (no Python required) from GitHub Releases.

Requirements

• Python 3.10+
• FFmpeg (brew install ffmpeg / apt install ffmpeg)
• At least one API key (OPENAI_API_KEY, ANTHROPIC_API_KEY, or GEMINI_API_KEY) or Ollama running locally

Output Structure

output/
├── manifest.json              # Single source of truth
├── transcript/
│   ├── transcript.json        # Full transcript with timestamps
│   ├── transcript.txt         # Plain text
│   └── transcript.srt         # Subtitles
├── frames/                    # Content frames (people filtered out)
├── diagrams/                  # Detected diagrams + mermaid code
├── captures/                  # Screengrab fallbacks
└── results/
    ├── analysis.md            # Markdown report
    ├── analysis.html          # HTML report
    ├── analysis.pdf           # PDF report
    ├── knowledge_graph.db     # SQLite knowledge graph
    ├── knowledge_graph.json   # JSON export
    ├── key_points.json        # Extracted key points
    └── action_items.json      # Tasks and follow-ups

Processing Depth

Documentation

Full documentation at planopticon.dev

License

MIT License -- Copyright (c) 2026 CONFLICT LLC