Start here
Connect Scrappycoco, choose a task, and get your first result.
Scrappycoco gives your AI agent or application one account and one interface
for tools from multiple providers. Use it to search the web, extract pages,
collect AI answers, and research social content. You choose what runs and can
see the provider, output, errors, and cost.
1. Connect your agent or application
For Codex, Claude Code, Cursor, or another agent with terminal access, paste:
Set up Scrappycoco: https://scrappycoco.ai/setup.mdFollow the browser sign-in prompt and wait for the agent to confirm setup.
See installation for manual and headless setup.
| Your environment | Start here |
|---|---|
| AI agent with terminal access | Install CLI + Skill |
| AI client with remote MCP support | Connect with MCP |
| Application backend, service, or CI | Use the REST API |
2. Try a task
Once connected, ask your agent:
Use Scrappycoco to extract the main text from https://example.com.
Inspect the available capability and provider options first. Return the text,
which provider ran, any missing content, and the reported cost.For requests you can run yourself, choose web content,
AI answers and citations, or
social research. See all use cases.
3. Check the result
Read the returned data and check whether it contains what you asked for.
A completed request can still contain missing, irrelevant, or truncated content.
Read results and costs explains the fields to check.
When you already know the provider and options, use Run. When you want to
test alternatives, Discover returns their results side by side so you or
your AI agent can choose. Compare and save configurations
walks through that process.
Updated about 1 month ago