tinyfish-web-agent_skill

This skill enables automated web data extraction and task automation using TinyFish, handling bot-protected sites and browser actions via natural language.
  • Shell

12

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill tinyfish-io/skills --skill tinyfish-web-agent

  • SKILL.md4.7 KB

Overview

This skill integrates the TinyFish web agent to extract data, scrape websites, and automate browser actions using natural-language goals. It requires a TinyFish API key and is designed for reliable, structured data output and bot-protected site handling. Use it to return JSON-formatted results directly from web pages or to run automated browser flows.

How this skill works

Before any call, the skill verifies the TINYFISH_API_KEY environment variable is set and stops if it is missing. It sends POST requests to TinyFish's run-sse endpoint with a URL and a natural-language goal that specifies the exact JSON structure to return. For bot-protected or geo-restricted sites, the request can include stealth browser profiles and proxy configuration. The final extracted data appears in the SSE stream's COMPLETE event in the resultJson field.

When to use it

  • You need structured JSON extraction from a single web page or list of items.
  • Scraping sites that use bot detection or require realistic browser behavior.
  • Automating sequences of browser actions and collecting the final outputs.
  • Extracting data from geo-restricted pages by routing through a country-specific proxy.
  • Running parallel, independent extractions for speed and reliability.

Best practices

  • Always run the pre-flight check to confirm TINYFISH_API_KEY is set; do not proceed without it.
  • Specify the exact JSON schema you want returned (fields and types) to get predictable output.
  • Make independent extraction tasks in parallel API calls rather than combining multiple sites in one call.
  • Use browser_profile: "stealth" for bot-protected sites and proxy_config for geographic targeting.
  • Read the SSE stream and use the event with type == "COMPLETE" and status == "COMPLETED" to retrieve resultJson.

Example use cases

  • Extract product lists and prices from an e-commerce category page as a JSON array.
  • Scrape job postings across multiple company pages in parallel and normalize fields.
  • Access a region-locked pricing page by routing through a US proxy and return structured pricing data.
  • Automate a login flow in stealth mode, navigate to a user dashboard, and extract account metrics.
  • Compare menu item prices between competitors by running separate, parallel extraction calls.

FAQ

The skill halts and asks you to set TINYFISH_API_KEY. You must provide the key before any API calls; do not fallback to other tools.

How do I handle bot-protected sites?

Include "browser_profile": "stealth" in the request body to use a stealth browser profile. Add proxy_config when you also need a specific country route.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational