List view
- No due date•6/6 issues closed
- No due date•4/4 issues closed
Goal: Agents work together intelligently without hardcoded sequences. - ✅ Implement CoordinatorAgent - ScraperAgent triggers AnalyzerAgent - AnalyzerAgent triggers ClassifierAgent - ClassifierAgent triggers ReporterAgent - ✅ Handle failures (retry, escalate) - ✅ Optional: Add simple UI or CLI to monitor agent activities
No due dateGoal: Create automatic summaries or community-facing reports. - ✅ Implement ReporterAgent - Summarize findings weekly or daily - Generate short readable reports - ✅ Format outputs in Markdown or simple HTML (for easy publication later) - ✅ Optional: Email notifications or Slack alerts for new findings
No due dateGoal: Extract structured metadata from detected surveillance-related documents. - ✅ Implement ClassifierAgent - Extract fields like: - Type of surveillance (CCTV, drones, facial recognition, etc.) - Location (city, neighborhood) - Entity responsible (government, private company) - ✅ Save structured metadata linked to original document - ✅ Improve prompting for extraction consistency
No due dateGoal: Analyze collected surveillance-related geospatial data to extract insights and flag relevant patterns. - ✅ Integrate local LLM (e.g., gemma2 via Ollama) with the agent - ✅ Design prompt templates for extracting structured surveillance metadata from raw tags - ✅ Implement AnalyzerAgent class with ability to: - Load raw Overpass JSON data from filesystem - Use LLM to normalize and extract fields: type, zone, mount, operator, public/private - Extract and normalize key fields (e.g., type, zone, mount type, coordinates, operator) - ✅ Identify and flag: - Cameras installed by law enforcement or government entities (e.g., "operator": "Polismyndigheten") - Surveillance zones of particular interest (e.g., "industrial", "town", "building") - Potential duplicates or overlapping coordinates - ✅ Summarize statistics: - Count cameras per zone type, mount type, or operator - Visualize spatial distribution (e.g., generate a GeoJSON or heatmap-ready dataset) - ✅ Store processed output in a structured format (e.g., JSON, CSV) - ✅ Log analysis steps and flagging outcomes
No due date•21/21 issues closedGoal: Enable your system to autonomously collect fresh documents. - ✅ Implement ScraperAgent - Scrape data from open street maps concerning surveillance using Overpass API - Download available data (json format) - ✅ Store raw collected data in filesystem - ✅ The agent remembers what has already been downloaded. Does not repeat downloads nor makes unnecessary requests to Overpass API - ✅ The agent flags requests that return empty
No due date•11/11 issues closedGoal: Have a runnable project with skeleton agent and basic tooling. ✅ Set up project structure (src/, agents/, tools/, memory/, etc.) ✅ Choose agent framework (CrewAI, AutoGen, or lightweight custom if you want full control) ✅ Integrate local LLM (Ollama, e.g., gemma2 or lightweight llama3) ✅ Implement minimal Agent base class ✅ Connect a lightweight database (e.g., SQLite or Redis) for storing collected data ✅ Create simple logging and config system
No due date•16/16 issues closed