Search documentation for Solo.io products using AI-powered semantic search.
The Solo.io - Internal Docs MCP Server enables your AI agents to search across the full range of Solo.io documentation and knowledge bases. Effortlessly find answers related to Istio, Ambient, kagent, and other Solo.io products.
To open Cursor and automatically add the MCP server, click install. Alternatively, add the following to your ~/.cursor/mcp.json file.
{
"mcpServers": {
"soloio-internal-docs-mcp": {
"url": "https://searchdocs.is.solo.io/mcp"
}
}
}To open VS Code and automatically add the MCP server, click install. Alternatively, add the following to your .vscode/mcp.json file in your workspace. To learn more, see the VS Code documentation.
{
"servers": {
"soloio-internal-docs-mcp": {
"type": "http",
"url": "https://searchdocs.is.solo.io/mcp"
}
}
}To add the MCP server to Claude Code, run the following command. To learn more, see the Claude Code documentation.
claude mcp add --transport http soloio-internal-docs-mcp https://searchdocs.is.solo.io/mcpAfter adding the server, you can verify it's connected:
claude /mcpAdd this MCP server to OpenCode by updating your OpenCode config file.
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"soloio-internal-docs-mcp": {
"type": "remote",
"url": "https://searchdocs.is.solo.io/mcp"
}
}
}If the server requires authentication, OpenCode will prompt you to authenticate when you first try to use it. If not, you can manually trigger the flow with opencode mcp auth <server-name>.
You can enable MCP servers on ChatGPT if you have a Pro, Plus, Business, Enterprise, or Education account. Follow the OpenAI documentation for instructions.
Use the following parameters when setting up your custom connector:
https://searchdocs.is.solo.io/mcp
This MCP server also works with OpenAI's Responses API when building autonomous agents.
Test the server easily using the MCP Inspector CLI tool. This is the quickest way to verify the server is working and see example results.
npx --yes @modelcontextprotocol/inspector --cli "https://searchdocs.is.solo.io/mcp" \
--transport http --method tools/call \
--tool-name search \
--tool-arg product=solo-enterprise-for-istio \
--tool-arg query='how to expose service across clusters' \
--tool-arg limit=1 | jq -r '.content[0].text'Here's an example of what the search tool returns:
## Result 1 (Score: 0.6277)
Product: solo-enterprise-for-istio
Version: latest
Section: Namespace sameness[](#namespace-sameness)
Hierarchy: Multi-cluster app routing > Namespace sameness[](#namespace-sameness)
Source: https://docs.solo.io/gloo-mesh/latest/ambient/multicluster/multi-apps/overview/
MCP is an open protocol supported by many clients. Your specific client documentation can advise you how to connect.
Use the server URL https://searchdocs.is.solo.io/mcp. No authentication is required.
Example configuration:
{
"soloio-internal-docs-mcp": {
"url": "https://searchdocs.is.solo.io/mcp"
}
}This server exposes the following MCP tools for AI agents to use:
Search documentation for the following products (Product (Aliases) - [collection source types]):
Usage Note: Evaluate search results: if they provide a complete and certain answer, then proceed. If the result is relevant but incomplete (e.g., ends mid - sentence), refers to a "see below" section, then use the get_chunks tool to get more context from the same document.
query
The search query to find relevant documentation (max 2000 characters)
product
The product to search documentation for
limit
Maximum number of results to return (default: 4)
type
Filter searched collections by Source Type. Use "all" (default) to query all data sources and return the most relevant results.
Retrieve sequential segments of a document to fill in missing context.
When to use: Use only when a search result is highly relevant but incomplete. This is the preferred way to "read the rest of the page" without performing additional keyword searches.
When to skip: Do not use if the search snippet already contains the full answer or if the search result is irrelevant to the user's query.
collection
The Qdrant collection name to query (e.g., 'istio-io_latest_docs')
url
The URL of the document to retrieve chunks from
startIndex
The starting chunk index (0-based, inclusive)
endIndex
The ending chunk index (0-based, inclusive). Will be clamped to the maximum available index if too large.
Retrieve the full markdown content of a documentation page by its URL from the database.
When to use: Use when you need the complete content of a specific page and chunk-based retrieval is insufficient.
Limits: Pages longer than 20,000 characters will return an error with the approximate token count. Set force: true to retrieve them anyway — note that very large pages consume significant context window space.
url
The exact URL of the documentation page to retrieve
force (optional)
Set to true to retrieve the page even if it exceeds the 20,000 character limit. Use with caution — large pages consume significant context.