Browser Automation

Automate exploratory QA of any web application with structured bug reports, screenshots, and videos. dogfood drives the agent-browser client to explore a target site, find visual, functional, UX, performance, console, and accessibility issues, and output a ready-to-share QA report with clear repro steps.

Automate existing Electron desktop apps like VS Code, Slack, Discord, Figma, Notion, and Spotify via agent-browser and Chrome DevTools Protocol (CDP). This skill helps you connect to a running Electron app, take snapshots, and interact with its UI as part of end-to-end desktop and workflow automation.

Automate Slack from the command line using browser automation. The slack skill connects to an existing Slack web session via agent-browser so you can check unread channels, scan DMs, search conversations, extract data, and capture structured reports as part of larger workflows.

Run agent-browser with headless Chrome inside Vercel Sandbox microVMs so Vercel-deployed apps can perform real browser automation, screenshots, and page interactions safely and at scale.

agent-browser lets AI agents control a Playwright-powered browser via inference.sh. Open pages, use @e element refs to click, type, drag, upload files, scrape content, and capture screenshots or video. Ideal for web automation, data extraction, and agent-driven browsing workflows.

agent-browser is a Chrome/Chromium automation CLI for AI agents and shell scripts. Use it to open pages, navigate, click, fill forms, capture snapshots, take screenshots, record video, profile performance, manage sessions, handle authentication, and automate end-to-end browser workflows.

agent-tools exposes the inference.sh CLI inside your agent so you can run 150+ AI apps from one place: image generation, video creation, LLMs, search, 3D, and Twitter automation. Ideal when you need a unified workflow runner for FLUX, Veo, Gemini, Grok, Claude, Seedance, OmniHuman, Tavily, Exa, OpenRouter, and more without managing GPUs or complex integrations.