Show HN: Building a no-code browser automation system for OSINT (github.com)

PyBA (Py Browser Automation) is a new no-code, LLM-powered browser automation framework in Python that turns a single natural-language prompt into human-like browser interactions. Built on Playwright, it autonomously navigates sites, fills forms, extracts data, performs OSINT tasks, and runs complex multi-step workflows. Key features include three reasoning/execution modes—Normal (deterministic), BFS (breadth-first for multiple success paths), and DFS (deep recursive exploration)—plus automatic Playwright trace generation for full reproducibility, per-step logging to local/server databases, exportable standalone Playwright scripts, and configurable anti-bot heuristics. It supports OpenAI, Vertex AI, and Gemini backends, fast social-media auth via environment variables, and parallel task execution. For AI/ML practitioners and security researchers, PyBA is significant because it embeds LLM “cognition” into end-to-end browser operations, enabling exploratory, human-like reasoning at scale while preserving auditability and replayability—critical for reproducible OSINT, automated testing, and research. The ability to persist every action/state and export deterministic scripts bridges exploratory LLM behavior with classical automation pipelines. Practical implications include faster investigative workflows, standardized logging for model-in-the-loop actions, and easier generation of training or evaluation datasets. As with any powerful automation/OSINT tool, responsible use and attention to legal and ethical constraints are essential to avoid misuse.