TinyFish is a CLI for web scraping and automation that escalates gracefully from simple fetches to full browser control. Start with search to find URLs, use fetch for clean content extraction from static pages, jump to agent when you need natural language instructions to click buttons or fill forms, and only reach for the CDP browser session when you need raw programmatic control. The agent command is the standout here: you pass a URL and a plain English goal like "extract all products as JSON" and it handles the interaction. Works well for bot-protected sites. The tool selection ladder (search → fetch → agent → browser) is smart, keeps costs down, and the parallel execution model means you fire off separate commands for independent sites rather than cramming everything into one brittle goal.
npx -y skills add tinyfish-io/tinyfish-cookbook --skill use-tinyfish --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills