[Feature Request] MCP Search Tool: Add pagination and filtering parameters

[Vigtu]

Hi team! 👋

Problem

Currently, the Mintlify MCP search tool (e.g., docs.agno.com/mcp) only accepts a single query parameter:

{
  "name": "SearchAgno",
  "parameters": {
    "query": {"type": "string"}
  }
}

When using AI coding assistants (Claude Code, Cursor, etc.) to query documentation, the search returns all results at once without any control over:

• Number of results
• Pagination for large result sets
• Filtering by content type or category

This leads to token inefficiency and context window pollution when working with large documentation sites.

Proposed Solution

Add optional parameters to the MCP search tool:

{
  "name": "SearchDocs",
  "parameters": {
    "query": {"type": "string", "required": true},
    "limit": {"type": "integer", "default": 10},
    "offset": {"type": "integer", "default": 0},
    "category": {"type": "string", "description": "Filter by doc category/group"}
  }
}

Use Case

As a developer using Claude Code with the Agno documentation MCP:

1. I search for "streaming agents"
2. Currently: Get all results (potentially 20+), consuming many tokens
3. With pagination: Get top 5 results first, request more only if needed

Additional Context

• The MCP specification (2025-06-18) already supports pagination for resources/list
• This would align the search tool with the spec's pagination patterns
• Similar to how GitHub's search APIs support per_page and page parameters

Thank you for considering! 🙏