Bret Bernhoft
A Linux Mint-based backend appliance for collecting, storing and analyzing Bluesky activity around selected keywords. The project is designed to run continuously as a headless collection node, gathering activity over time from the ATProto ecosystem and preserving them in a local SQLite database for later analysis.
At its core, the system polls the ATProto API for chosen topics, processes the returned posts and stores useful information such as text content, handles, engagement, keywords and response data. Each run is logged so the appliance can track not only what it collected, but also how reliably the device operated. This includes timestamps, success or failure states, duplicates, inserted records, total volume and errors.
The goal is to build a small, reliable social signal collection appliance operating independently on a local network. Rather than depending on an external analytics platform, the project emphasizes simplicity and transparency; a Linux Mint machine, Flask API, SQLite database and scheduled collection.
This project is especially useful for long-term observation of public conversation patterns. By collecting posts containing particular keywords over months, the system can support later analysis of topics, authors, engagement, trends, health and changes in public interest. The stored data will eventually be exported into reports and frontend D3.js graphics.
The project is intentionally built in phases. The first phase focuses on data collection and storage. The second phase adds automation, uptime tracking and regular report generation. Later phases include dashboards, trend detection, comparisons, frequency analysis and data exports.
Mark Pedersen
Tobias
C. M. Herron