# lyrenth/lyrenth-mcp [Health: Active]

**Category:** 📂 Browser Automation  
**Repository:** https://github.com/lyrenth/lyrenth-mcp  
**GitHub Stars:** 1  
**npm Downloads (last month):** 725  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/lyrenth-lyrenth-mcp

## Description
Read any URL as a clean AIDocument (Markdown + structure) through Lyrenth's cached index.

## Tools
Capabilities this server exposes over MCP:

- **read_url** — The page as a clean AIDocument: a short provenance header (token count + how much smaller than raw HTML) plus the Markdown body. `fresh: true` forces a fresh fetch instead of the cached copy; `max_tokens` caps the body to your context budget, trimmed at a clean paragraph or sentence boundary.
- **read_urls** — Up to 20 pages in one call, each as a clean AIDocument, with per-URL error isolation (a failed URL is reported and doesn't block the others). Billed one credit per successfully-read URL.
- **check_usage** — Your plan tier, credits used against your monthly limit, credits remaining, and the reset date.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "lyrenth-mcp": {
    "command": "npx",
    "args": ["-y","path"],
    "env": {
      "LYRENTH_API_KEY": "",
      "LYRENTH_API_URL": ""
    }
  }
}
```

**Requires environment variables:** `LYRENTH_API_KEY`, `LYRENTH_API_URL` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What lyrenth/lyrenth-mcp MCP server does

lyrenth/lyrenth-mcp MCP server converts public web pages into AIDocuments for MCP clients. Each result contains a cleaned Markdown representation with the page title, description, and structure, while removing navigation and other boilerplate. A provenance header reports the document's token count and its reduction compared with the source HTML.

The server reads through Lyrenth's cross-caller cache. For domains verified with Lyrenth, the returned document may use the publisher's canonical version. This makes the server suited to agents that need readable web content without passing complete HTML into the model.

## How it works

The server supports two transports. A hosted MCP endpoint is available at `https://api.lyrenth.com/mcp` for clients that can connect to remote HTTP servers and send a custom authorization header. A local stdio process can be started with the npm package and an API key in its environment.

`read_url` accepts one URL and can optionally request a fresh fetch or limit the response body with `max_tokens`. Token-limited output ends at a paragraph or sentence boundary. `read_urls` accepts between one and 20 URLs in a single call. Each URL is handled independently, so one failure is reported without preventing successful results for the other URLs. Successful reads consume one credit per URL.

## Setup and configuration

lyrenth/lyrenth-mcp MCP server requires a Lyrenth API key. The README directs users to obtain a free key with a stated allowance of 2,000 reads per month and no card. For a local setup, configure `LYRENTH_API_KEY` and run `npx -y lyrenth-mcp`; the same command can be added to an MCP client configuration.

`LYRENTH_API_URL` is optional and defaults to `https://api.lyrenth.com`. It can point the client at a staging or self-hosted API. Claude Desktop can also install the supplied `.mcpb` release bundle, which asks for the key during setup. Remote clients use the hosted URL with `Authorization: Bearer <key>`.

## Tools and capabilities

- `read_url`: Returns one cleaned AIDocument, with optional cache bypass and token limiting.
- `read_urls`: Reads up to 20 URLs and isolates errors per URL.
- `check_usage`: Reports the plan tier, monthly credits used, remaining credits, and reset date.
- MCP tool annotations declare read-only, non-destructive, idempotent, and open- or closed-world properties as explicit booleans.

## Limitations and notes

lyrenth/lyrenth-mcp MCP server requires an API key and usage is credit-metered. Cached content may not be the newest available page unless `fresh: true` is supplied. Batch requests are limited to 20 URLs, and `max_tokens` limits the returned body rather than guaranteeing a particular amount of source content.

The local process sends requested URLs and the API key to Lyrenth's API. The README states that local page content, conversation context, and telemetry are not sent. Lyrenth's handling of fetched pages and account data is covered by its privacy policy. The project is licensed under MIT.

_Full upstream README: https://allmcps.com/mcp/lyrenth-lyrenth-mcp/readme_