Installation and sign-in

Node, npm, or npx is missing

Paste Set up Scrappycoco: https://scrappycoco.ai/setup.md into your agent.
The managed installer provides its own private runtime; you do not need to
install Node separately. Use the full launcher path printed by setup when your
current agent cannot find scrappycoco on PATH. The installer does not edit shell
startup files.

For all commands below, use that same working CLI invocation. An MCP-only client
should use its native OAuth connection and tools instead of these CLI commands.

Setup stopped after the browser callback

Return to the terminal that started setup. The callback only means the browser
sent authorization; setup is complete after the terminal confirms credential
storage and catalog verification.

Run:

scrappycoco doctor --json

The browser cannot reach localhost

Use the manual flow on remote or headless hosts:

scrappycoco setup --no-browser --manual-callback

Keep the command running and paste the final callback URL into that terminal.
Do not share or save the callback URL. Start a new setup command if the old one
timed out.

The agent cannot see the skill

Check the installed target:

scrappycoco skill status --json

Ask the agent to read the SKILL.md at target in that response (or skill.path
in the setup result), then apply it in the current conversation. The CLI works
immediately; restarting is not a prerequisite for using it. Automatic skill
discovery for future sessions depends on the client. Only follow a reload
instruction when the client actually reports that it needs one for discovery.

The MCP tools are missing

Confirm all three stages:

The plugin or remote server definition is installed and trusted.
The current client session loaded it.
OAuth for https://api.scrappycoco.ai/mcp completed.

For Claude Code, run /reload-plugins after a mid-session installation, then
use /mcp to authenticate. Verify that list_scrapers is available.