Connect your AI to OyaPilot.

OyaPilot runs a read-only MCP server so your own AI, Claude Code, Claude Desktop, Cursor, or any MCP-compatible client, can look up your jobs, applications, and queue directly. You mint a personal token, paste it into your client, and your AI can answer questions like "what is in my queue today" or "find me remote staff engineer roles" using your real OyaPilot data.

Connect URL

https://oyapilot.ai/api/mcp

Streamable HTTP transport. Every request needs an Authorization header with your personal token.

1. Create a personal token

Sign in to OyaPilot, open Settings > Connected AIs, and create a token. It is shown to you exactly once, so copy it before you close the box. OyaPilot only ever stores a one-way hash of it, never the token itself, and you can revoke it anytime from the same page.

Open Connected AIs settings

2. Set up your client

Replace YOUR_TOKEN below with the token you copied.

Claude Code

claude mcp add --transport http oyapilot https://oyapilot.ai/api/mcp \
  --header "Authorization: Bearer YOUR_TOKEN"

Remove it later with claude mcp remove oyapilot.

Claude Desktop

Edit your claude_desktop_config.json and add:

{
  "mcpServers": {
    "oyapilot": {
      "type": "http",
      "url": "https://oyapilot.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Fully quit and reopen Claude Desktop after saving.

Cursor

Add this to .cursor/mcp.json (or ~/.cursor/mcp.json to make it available in every project):

{
  "mcpServers": {
    "oyapilot": {
      "url": "https://oyapilot.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Reload Cursor after saving.

claude.ai web connectors and ChatGPT: these require an OAuth connection flow rather than a pasted personal token. That is coming in a later phase; for now, personal tokens work with Claude Code, Claude Desktop, Cursor, and any other MCP client that accepts a custom header on a remote HTTP server.

Read-only, and here is what it cannot do

Every tool a connected AI can call is a read. None of them write, submit, or send anything. Specifically, a connected AI cannot:

  • Submit or apply to any job on your behalf.
  • Send email from your account.
  • Change any account or agent setting.
  • Access your account password or payment details.
  • Read your full résumé file or profile beyond a completeness and contact summary.
  • See message bodies in your mail threads, only subject and sender metadata.

A token is capped at 2,000 tool calls a day and throttles automatically past that. You can revoke it in one click from Settings > Connected AIs, which invalidates it immediately for every client using it.

Tool reference

search_jobsSearch verified, non-ghost-filtered job postings by keyword, location, and remote status.
get_jobGet full detail for one job posting by its source ATS and external id.
ghost_checkRun a ghost-job risk read on pasted posting text or a posting URL; returns a risk band and reasons.
list_applicationsList your tracked applications, newest first, optionally filtered by status.
get_applicationGet one of your tracked applications by id.
list_saved_jobsList your saved (bookmarked) jobs.
get_queueGet your morning queue: items ready for one-tap approval and items that need you.
list_mail_threadsList metadata for your mail threads (subject, sender, category), newest first. Never message bodies.
get_profile_summaryA lean, non-sensitive summary of your Application Profile: completeness, résumé presence, a few fields.
Create your token How we protect your data →