Skip to main content
Claude Code is Anthropic’s coding agent for the terminal. With Spyro AI connected, it can check your audit issues, rankings and content calendar while you work on your site’s code.

Connect with sign-in

1

Add the server

2

Sign in

Start Claude Code and run /mcp. Pick spyro-ai and sign in in your browser. Choose the workspaces and the access level, then click Allow.

Connect with a key

Use a key for scripts, CI or machines without a browser. Create one in Settings → MCP. See Manage keys and apps.

Choose where it is saved

Claude Code can store the server at three levels with --scope: For a shared .mcp.json, never commit a key. Use an environment variable that each person sets:
.mcp.json

Manage it

Inside Claude Code, /mcp shows status and lets you sign in again.

Try these prompts

  • “Use Spyro AI to list my critical audit issues, then find the files in this repo that cause them.”
  • “Check Spyro AI for pages with a low click rate and rewrite their title tags.”
  • “Use Spyro AI to audit https://example.com/pricing and apply the top fixes to our pricing page.”
  • “Which of my pages are losing traffic in Search Console? Suggest code changes.”

Tips

  • Pair audit and Search Console data with code edits. This is where a coding agent adds the most.
  • Keep the connection on Read. If you allow actions, keep Claude Code’s tool approval on. Each change also waits for you to confirm it in Spyro AI.
  • Run /mcp__spyro-ai__fix-audit, /mcp__spyro-ai__weekly-report or /mcp__spyro-ai__publish-article for ready-made workflows (the middle part is the server name you chose). See Prompts for coding agents.
  • One key per machine or pipeline, with a low daily cap, so you can revoke it alone.

If it fails

  • 401 error: the key is wrong or revoked, or you need to sign in again with /mcp.
  • Server shows as failed: run claude mcp get spyro-ai and check the address.
  • More: Troubleshooting.