Detectivision runs a public Model Context Protocol server with this documentation. Connect it to your coding agent (Claude Code, Cursor, VS Code, Windsurf and others) and the agent can search and read the docs and the OpenAPI description while it writes your integration, instead of guessing.
| URL | https://detectivision.ai/mcp |
| Transport | Streamable HTTP |
| Authentication | None. The server only serves public documentation. |
| Access | Read-only. It never calls the API or spends credits. |
Connect your agent#
claude mcp add --transport http detectivision https://detectivision.ai/mcp// ~/.cursor/mcp.json (or .cursor/mcp.json in your project)
{
"mcpServers": {
"detectivision": {
"url": "https://detectivision.ai/mcp"
}
}
}// .vscode/mcp.json
{
"servers": {
"detectivision": {
"type": "http",
"url": "https://detectivision.ai/mcp"
}
}
}// ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"detectivision": {
"serverUrl": "https://detectivision.ai/mcp"
}
}
}// Settings → Connectors → Add custom connector, or in claude_desktop_config.json:
{
"mcpServers": {
"detectivision": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://detectivision.ai/mcp"]
}
}
}Any other client that supports remote MCP servers over Streamable HTTP works with the same URL.
Tools#
| Tool | Arguments | What it returns |
|---|---|---|
search_docs |
query (string), limit (1–20, default 8) |
The sections that best match, with the page path to read next. |
get_doc |
path (a path such as quickstart or api/createModeration, or a docs URL) |
The whole page as Markdown. |
list_docs |
none | Every page with its path, group and summary. |
get_openapi |
none | The OpenAPI 3.1 description of the API as JSON. |
Try it#
Ask your agent something like:
- "Using the Detectivision docs, write a Python function that checks an image URL and returns the flags it raised."
- "How does Detectivision bill live stream sessions? Check the docs."
- "Add Detectivision text moderation to this Express route, with retries on 503."
Tip
The server always serves the current documentation: it is published together with this site.
Without MCP#
The same content is available as plain files, for tools that read the web:
/llms.txt: an index of every page, with links to their Markdown./llms-full.txt: all pages in one file.- Any page as Markdown: add
.mdto its address, for example/docs/quickstart.md. /openapi.json: the API description.
See llms.txt for details.