Lovely Router MCP server
com.lovelyrouter/mcplovelyrouter.comstreamable-httpv1.0.0Scanned
One MCP URL for web search, page reading, pay-per-result data tools and a browser agent.
13 tools: 1 do something that can't be undone, 4 change things and 8 only read.
- Reads private data · 1
- Reads text others write · 4
- Can send data out · 2
- No findings
Tools listed without sign-in. Remote posture
With Anlyon
1 tool here does something that can't be undone. Govern it.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the Lovely Router MCP server can do
The Lovely Router MCP server is published by lovelyrouter.com, a domain the MCP Registry verified. It has 13 tools. It can do things that cannot be undone, such as sending, paying or publishing (browser_reply) and create or change data (4 tools: generate_image, run_tool, browser_run and browser_takeover), but cannot delete data. 8 tools only read.
1 tool reads private data (web_search); 4 tools bring in text other people can write (web_search, web_fetch, search_tools and 1 more); 2 tools can send data out of your system (web_fetch and browser_reply). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
None of the fixed checks matched anything in its tool definitions. That is not a safety guarantee.
Its remote endpoint lists its tools without sign-in.
Last scanned Nov 17, 2026, version 1.0.0.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
13 tools
Can't be undone 1Claude Code: ask
browser_replyReply to a browser run waiting for a human: {"answer": "…"} for a question, {"approve": true|false} for an approval, {"done": true} after the user finished a takeover.Can send data outNo findings
Why it is in this group
- The name "browser_reply" says "reply": a word for sending, and what is sent cannot be taken back.
What it can reach
- outboundThe name "browser_reply" says "reply": it sends or publishes something to other people.
Description, as the server wrote it Reply to a browser run waiting for a human: {"answer": "…"} for a question, {"approve": true|false} for an approval, {"done": true} after the user finished a takeover. Codes and passwords the user gives should be sent by them through the API "secret" reply, not through this tool.
Changes things 4Claude Code: ask
browser_runRun an AI browser agent on a website you have to interact with: log in, click through, fill and submit forms, multi-page flows.No findings
Why it is in this group
- The name "browser_run" says "run": a word for a change.
Description, as the server wrote it Run an AI browser agent on a website you have to interact with: log in, click through, fill and submit forms, multi-page flows. To read what a page says use web_fetch, to find pages use web_search: both are faster and cheaper. The browser runs in the cloud and cannot open localhost or private addresses. Give a goal (with every value it needs) and a start URL; it navigates, fills forms and reports back. It can log in with secrets the user saved in their account vault (pass their names — values never reach any model), resume a saved session (profile), and pause to ask a question, hand the browser to a human (CAPTCHA) or get approval before a purchase. Billed per step. Returns the run after up to ~45s; poll with browser_status.
browser_takeoverAsk a browser run to hand the browser to the user now; the run status then carries a live_view_url.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Description, as the server wrote it Ask a browser run to hand the browser to the user now; the run status then carries a live_view_url.
generate_imageCreate an image from a description with GPT Image 2.5 (GPT Image 2 as backup): illustrations, photos, product shots, posters, icons, diagrams with little text, UI mockups.No findings
Why it is in this group
- The name "generate_image" says "generate": a word for a change.
Description, as the server wrote it Create an image from a description with GPT Image 2.5 (GPT Image 2 as backup): illustrations, photos, product shots, posters, icons, diagrams with little text, UI mockups. Give reference images in `images` to edit a picture or to keep a product, logo, person or style the same. $0.04 per image, any size, with or without references; nothing if it fails. It takes 30 s to 2 min. The answer is a download link that works for 24 hours without a key: save the file with the curl command it gives. If the image is not ready within the wait, call get_image with the job id (do not generate again: that is a second image, billed again).
run_toolRun a data tool (an id from suggest_tools or search_tools) and get its results.No findings
Why it is in this group
- The name "run_tool" says "run": a word for a change.
Description, as the server wrote it Run a data tool (an id from suggest_tools or search_tools) and get its results. `input` must follow the tool's input_schema: start from suggest_tools' example_input or get_tool's example and change only the values. It waits for the results; a longer job answers with a run_id — then call get_run with it (do not run the tool again). Results come back compact (empty fields dropped, long text cut, about 12,000 characters at most): pass `fields` to keep only the keys you need, and call get_run with `offset` for more items. Billed per call + per result, capped by max_results: ask for only as many results as the task needs. Never run a tool again to see other fields or more of the same results: get_run re-reads a finished run for free.
Only reads 8Show 8 read-only tools
browser_statusStatus of a browser run (waits up to ~45s for progress).Reads text others writeNo findings
Why it is in this group
- The name "browser_status" says "status": a word for reading.
What it can reach
- untrusted-inputThe name "browser_status" says "status" and "browser": it reads web pages, which anyone can write.
- untrusted-inputThe description's first sentence says "status" and "browser": it reads web pages, which anyone can write.
Description, as the server wrote it Status of a browser run (waits up to ~45s for progress). When status is waiting_for_human, "needs" says what it needs: a question to answer, an approval, or a takeover (live_view_url for the user).
get_imageAn image from generate_image that was still being made: waits for it, then returns its download link.No findings
Why it is in this group
- The name "get_image" says "get": a word for reading.
Description, as the server wrote it An image from generate_image that was still being made: waits for it, then returns its download link. Free.
get_runStatus and results of a tool run: waits for it to finish, then returns its results compact.No findings
Why it is in this group
- The name "get_run" says "get": a word for reading.
Description, as the server wrote it Status and results of a tool run: waits for it to finish, then returns its results compact. `offset` pages through the items; `fields` keeps only the keys you need. Free: use it to re-read a finished run with other fields instead of running the tool again.
get_toolA tool's input schema (field descriptions carry its query syntax), an example input and its price.No findings
Why it is in this group
- The name "get_tool" says "get": a word for reading.
Description, as the server wrote it A tool's input schema (field descriptions carry its query syntax), an example input and its price. Read it before run_tool unless suggest_tools already gave you example_input: run_tool rejects input that doesn't match the schema.
search_toolsFind data tools for a platform by describing what you need (Google Maps, TikTok, Instagram, LinkedIn, Amazon, Reddit, X, YouTube, reviews, jobs…).Reads text others writeNo findings
Why it is in this group
- The name "search_tools" says "search": a word for reading.
What it can reach
- untrusted-inputThe description's first sentence says "find" and "review": it reads comments, which other people can write.
Description, as the server wrote it Find data tools for a platform by describing what you need (Google Maps, TikTok, Instagram, LinkedIn, Amazon, Reddit, X, YouTube, reviews, jobs…). Returns tool ids with prices. For plain web search use web_search.
suggest_toolsDescribe the whole task (e.g.No findings
Why it is in this group
- The description's first verb is "describe": a word for reading.
Description, as the server wrote it Describe the whole task (e.g. "find vegan restaurants in Austin and summarise their latest reviews", "get the transcript and top comments of this YouTube video") and get an ordered plan in ~0.5s: the tool to run first, an optional second tool, close alternatives, each with an example input to adapt (see input_required) and the estimated cost. Free. Use it first whenever you need data from YouTube, Reddit, X, TikTok, Instagram, LinkedIn, Google Maps, Amazon, job boards or similar sites: a data tool returns it structured in one call, much faster than fetching or scraping their pages.
web_fetchRead one web page as clean Markdown (title and text, no menus or ads).Reads text others writeCan send data outNo findings
Why it is in this group
- The name "web_fetch" says "fetch": a word for reading.
What it can reach
- untrusted-inputThe name "web_fetch" says "fetch" and "web": it reads web pages, which anyone can write.
- untrusted-inputThe description's first sentence says "read" and "web": it reads web pages, which anyone can write.
- untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Claude Code: ask, because it is tagged outbound.
Description, as the server wrote it Read one web page as clean Markdown (title and text, no menus or ads). Always use this, not browser_run, when you only need what a page says. For YouTube, Reddit, X, TikTok, Instagram, LinkedIn, Google Maps or Amazon pages use suggest_tools and run_tool instead: they return the data itself (video details, transcripts, comments, posts, reviews, prices) structured and complete. $0.0009 per page, nothing if it cannot be read. Long pages come in parts: to read on, call again with the run_id and offset from the answer (free). It cannot open localhost, private addresses or files in your sandbox.
web_searchSearch the web (pages, news, docs, blogs) and get up to 50 results with title, URL, date and a summary written for you to read.Reads private dataReads text others writeNo findings
Why it is in this group
- The name "web_search" says "search": a word for reading.
What it can reach
- private-dataThe description's first sentence says "search" and "doc": it reads documents.
- untrusted-inputThe name "web_search" says "search" and "web": it reads web pages, which anyone can write.
- untrusted-inputThe description's first sentence says "search" and "web": it reads web pages, which anyone can write.
Description, as the server wrote it Search the web (pages, news, docs, blogs) and get up to 50 results with title, URL, date and a summary written for you to read. Use it first whenever you need to look something up: one call, $0.0009 per search, nothing when there are no results. Query operators: site:example.com (several: site:a.com OR site:b.com), -site:example.com, "exact phrase", -word, lang:en (or lang:zh,en), after:2026-01-01, before:2026-06. For Reddit, X/Twitter, YouTube, Google Maps, Amazon, TikTok and other platforms use suggest_tools and run_tool instead.
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
8 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 private-data tool breaks every one.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write4
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out2
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
{
"permissions": {
"deny": [
"mcp__lovelyrouter__web_search"
]
}
}Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 17, 2026, 15:19 UTC, unchanged since Nov 17, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- yes
- Tools that change things, listed without credentials
- 5
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- not observed
- Authorization servers
- not observed
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- not observed
- Client ID metadata documents
- not observed
- HighThe whole server
The endpoint served tools/list without credentials, and the list has 5 tools that change things (1 irreversible, 4 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://lovelyrouter.com/mcp: tools/list returned 13 tools to an anonymous client, including browser_reply (irreversible), generate_image (write), run_tool (write), browser_run (write), browser_takeover (write)
Findings
Matches of fixed checks, with the text they matched.
No check matched anything in these tool definitions. That is not a safety guarantee.
History
Changes to its tool definitions and to its remote endpoint, newest first.
No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.
Install
Pinned to version 1.0.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http lovelyrouter https://lovelyrouter.com/mcp --header 'Authorization: <your-authorization>'claude mcp add --transport http lovelyrouter https://lovelyrouter.com/mcp --header 'Authorization: <your-authorization>'Run in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__lovelyrouter__web_search",
"mcp__lovelyrouter__get_image",
"mcp__lovelyrouter__search_tools",
"mcp__lovelyrouter__suggest_tools",
"mcp__lovelyrouter__get_tool",
"mcp__lovelyrouter__get_run",
"mcp__lovelyrouter__browser_status"
],
"ask": [
"mcp__lovelyrouter__web_fetch",
"mcp__lovelyrouter__generate_image",
"mcp__lovelyrouter__run_tool",
"mcp__lovelyrouter__browser_run",
"mcp__lovelyrouter__browser_reply",
"mcp__lovelyrouter__browser_takeover"
],
"deny": []
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as lovelyrouter. Review it before you use it.
Capability tags behind 5 rules
- mcp__lovelyrouter__web_search private-data, untrusted-input
- mcp__lovelyrouter__web_fetch untrusted-input, outbound
- mcp__lovelyrouter__search_tools untrusted-input
- mcp__lovelyrouter__browser_status untrusted-input
- mcp__lovelyrouter__browser_reply outbound
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/com.lovelyrouter/mcp/verify?hash=443e1cce117fa2d51751cdbe5b02354368575e3ab46ead5f056c9f49f34b95a2" | jq -e '.changed == false'Pinned hash 443e1cce117f, from Nov 17, 2026, 04:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/com.lovelyrouter/mcp/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/com.lovelyrouter/mcp)Source: MCP Registry. Synced Nov 17, 2026, 04:17 UTC.
Lovely Router MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Lovely Router MCP server safe?
- No scan can promise that. What its definitions show: 13 tools, 5 of which can change, send or delete something, and 0 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the Lovely Router MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the Lovely Router MCP server do things that cannot be undone?
- Yes. 1 tool can send, pay, publish or merge: browser_reply.
What tools does the Lovely Router MCP server have?
- 13 tools: 1 irreversible, 4 write, 8 read. browser_reply, generate_image, run_tool, browser_run, browser_takeover, web_search, web_fetch, get_image and 5 more.
Does the Lovely Router MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the Lovely Router MCP server send data to outside services?
- 2 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): web_fetch and browser_reply.
Has the Lovely Router MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 17, 2026.