Vibbly Agent

Use this skill when a user asks an agent to search podcasts, retrieve existing transcripts, or transcribe podcast, audio, or video content through Vibbly.

Connection

• Prefer the Vibbly MCP server when available.
• MCP endpoint: https://usevibbly.com/mcp
• Connect through an OAuth-capable MCP client. The client should open a Vibbly browser login and grant page.

Example remote MCP config:

{
    "mcpServers": {
        "vibbly": {
            "type": "http",
            "url": "https://usevibbly.com/mcp"
        }
    }
}

For clients that only support local command MCP servers, bridge with pnpm dlx:

{
    "mcpServers": {
        "vibbly": {
            "command": "pnpm",
            "args": ["dlx", "mcp-remote", "https://usevibbly.com/mcp"]
        }
    }
}

Tool Policy

• Use search_podcasts for podcast catalog search.
• Use search_episodes when the user asks for episodes, topics, titles, or audio URLs.
• Use get_podcast_episodes after identifying a podcast.
• Use retrieve_transcript before creating any new transcription. It can read public podcast transcripts and private transcripts owned by the signed-in Vibbly account.
• Use transcribe_media_url only when the user explicitly asked to transcribe, or after asking for confirmation, because new transcription work spends the user's Vibbly credits.
• Include duration_seconds in transcribe_media_url when known so Vibbly can pre-check credit balance.

User-Facing Behavior

• Be explicit when an action can spend credits.
• Prefer existing transcripts before new transcription.
• Treat agent transcript access as the same as the signed-in user's Vibbly access.
• For long media, mention that Vibbly charges by transcription minutes.
• Return transcript IDs and Vibbly URLs when tools provide them so the user can inspect the result.