sidebutton/sidebutton

Sidebutton is a classic 'feature-as-a-product' that is currently being squeezed by both infrastructure providers and frontier labs. While it correctly identifies the Model Context Protocol (MCP) as a vital integration point for agents, its core offering—browser automation—is a highly commoditized space. With only 5 stars and 0 forks over 100+ days, the project lacks the community momentum needed to build a data or workflow moat. Competitively, it faces three massive headwinds: 1. Frontier Labs: OpenAI (Operator), Anthropic (Computer Use), and Google (Jarvis) are building native, model-level browser interaction that bypasses the need for middle-tier YAML wrappers. 2. Established Startups: Companies like MultiOn, Skyvern, and LaVague have raised significant capital and possess much deeper automation stacks (including self-healing selectors and computer vision-based navigation). 3. Infrastructure Standards: Microsoft’s Playwright and the browser vendors themselves are increasingly building 'agent-ready' modes. The 'Knowledge Packs' concept is the only potential path to defensibility, but without a massive library of pre-built, high-value workflows for enterprise apps (SAP, Salesforce, etc.), it remains a thin wrapper around standard automation libraries. Platform domination risk is high because browser control is a natural extension of both the OS and the LLM provider.