Schema | MCP Web Fetch Server

MCP Web Fetch Server

Describes the environment variables required to run the server.

Name	Required	Description
`MCP_AUTH_TOKEN`	No	Bearer token for HTTP mode authentication and dashboard auth
`FETCH_SEARXNG_URL`	No	URL of SearXNG instance for web search fallback (default http://localhost:8080)
`FETCH_ADMIN_ENABLED`	No	Set to 'false' to disable the management GUI (default true)
`FETCH_LOCAL_FILES_ROOT`	No	Root directory for sandboxed local file access (opt-in, disabled by default)

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Features and capabilities supported by this server

Protocol revision2025-11-25

Capability	Details
`tools`	{ "listChanged": false }
`prompts`	{ "listChanged": false }
`resources`	{ "subscribe": false, "listChanged": false }
`completions`	{}
`experimental`	{}

Functions exposed to the LLM to take actions

Name	Description
fetch_urlA	Fetch a public HTTP/HTTPS URL and return page content as markdown. Use start_index to read long pages in chunks.
fetch_metadata_toolA	Return HTTP metadata for a URL using a HEAD request.
batch_fetchA	Fetch up to 10 public URLs concurrently and return each page's content (or error) in one response.
web_searchA	Search the web via DuckDuckGo (no API key required) and return titles, URLs, and snippets for the top matches. Automatically falls back to a local SearXNG instance (FETCH_SEARXNG_URL) if DuckDuckGo's scrape fails.
extract_linksB	Fetch a URL and extract all links and images from the page as a list.
summarize_urlA	Fetch a URL and ask the connected client's LLM (via MCP sampling) to summarize it. Requires a client that supports the sampling capability; otherwise use fetch_url and summarize the content yourself.
read_fileA	Read a text file from an allowed local directory (configured via FETCH_LOCAL_FILES_ROOT, plus any directories the client exposes as roots). Path is relative to the allowed directory.
list_dirA	List the contents of a directory within an allowed local directory. Leave path empty to list the root itself.
write_fileA	Write text content to a file within an allowed local directory. If the file already exists and overwrite=false, the user will be asked to confirm before it is replaced.

Interactive templates invoked by user choice

Name	Description
`fetch`	Fetch a URL and extract its contents as markdown.
`research_topic`	Guides the assistant through searching, fetching, and synthesizing information about a topic using this server's tools.
`summarize_page`	Fetch a URL and produce a concise, faithful summary of its content.
`extract_key_facts`	Fetch a URL and pull out structured facts (names, dates, numbers, claims).
`compare_sources`	Fetch multiple URLs and compare/contrast what they say about a shared claim or question.

Contextual data attached and managed by the client

Name	Description
`server-settings`	Current (secret-redacted) server configuration and version.
`fetch-history`	The most recent URLs fetched by this server instance, most recent first. Successfully fetched pages may also be readable via the fetch-cache:// resource template.

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/narimanamiri/mcp-fetch-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server