Agentwatch MCP server
lighting.comiza/agentwatchcomiza.lightingstreamable-httpv1.0.0Scanned
Check if ChatGPT, Claude and Perplexity can reach, read and quote a website.
4 tools: 2 change things and 2 only read.
- Reads private data · 1
- Reads text others write · 1
- Can send data out · 2
- No findings
Tools listed without sign-in. Remote posture
With Anlyon
2 tools here change things. Govern them.
- 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 Agentwatch MCP server can do
The Agentwatch MCP server is published by comiza.lighting, a domain the MCP Registry verified. It has 4 tools. It can create or change data (2 tools: verify_crawler_visit and verify_signed_request), but cannot delete data. 2 tools only read.
1 tool reads private data (check_site); 1 tool brings in text other people can write (check_site); 2 tools can send data out of your system (check_site and verify_signed_request). 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 18, 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.
4 tools
Changes things 2Claude Code: ask
verify_crawler_visitGiven the IP address and User-Agent from a log line, say whether that visit really came from the crawler it claims to be.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 Given the IP address and User-Agent from a log line, say whether that visit really came from the crawler it claims to be. Checks the address against the vendor's own published ranges where they exist, and falls back to forward-confirmed reverse DNS. On one ordinary server's logs, 14% of requests claiming to be Googlebot were not Google. Use this before anyone writes a firewall rule based on a user agent string, because that string is a claim and not proof.
verify_signed_requestVerify a Web Bot Auth signature (RFC 9421, Ed25519) on a captured request.Can send data outNo findings
Why it is in this group
- The name "verify_signed_request" says "request": a word for a change.
What it can reach
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Verify a Web Bot Auth signature (RFC 9421, Ed25519) on a captured request. Unlike an address check, this proves who sent that exact request: it fetches the signer's published key directory, rebuilds the signature base from the covered components, and checks the signature. Use it when a request carries Signature-Input and Signature headers. Returns 'unsigned' when they are absent, and 'cannot_verify' rather than 'invalid' whenever the limitation is on our side.
Only reads 2Show 2 read-only tools
check_siteFetch a website the way ChatGPT, Claude, Perplexity, Google and Copilot fetch it, and report what stops them.Reads private dataReads text others writeCan send data outNo findings
Why it is in this group
- The description's first verb is "fetch": a word for reading.
What it can reach
- private-dataThe description's first sentence says "fetch" and "chat": it reads messages.
- untrusted-inputThe description's first sentence says "fetch" and "website": 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 Fetch a website the way ChatGPT, Claude, Perplexity, Google and Copilot fetch it, and report what stops them. Checks robots.txt per crawler with the deciding line number, makes real requests as each crawler to catch firewall and CDN rules that robots.txt cannot express, and measures whether the page carries readable text without JavaScript. Use this before advising anyone on why their site does not appear in AI answers: the usual cause is one of these three and none of them are visible from the outside.
list_ai_crawlersThe robots.txt tokens used by the major assistants, separated by purpose: the ones that build the index answers come from, the ones that fetch a page live when a person asks, and t…No findings
Why it is in this group
- The name "list_ai_crawlers" says "list": a word for reading.
Description, as the server wrote it The robots.txt tokens used by the major assistants, separated by purpose: the ones that build the index answers come from, the ones that fetch a page live when a person asks, and the ones that only collect training data. Blocking the last group costs no visibility; blocking the first two removes a site from the answers. Use this before writing or editing a robots.txt.
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.
2 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 untrusted-input tool breaks every one.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write1
- 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__agentwatch__check_site"
]
}
}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 20, 2026, 19:15 UTC, unchanged since Oct 22, 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
- 2
- 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
- WarnThe whole server
The endpoint served tools/list without credentials, and the list has 2 tools that change things (2 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://comiza.lighting/agentwatch/mcp: tools/list returned 4 tools to an anonymous client, including verify_crawler_visit (write), verify_signed_request (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 agentwatch https://comiza.lighting/agentwatch/mcpclaude mcp add --transport http agentwatch https://comiza.lighting/agentwatch/mcpRun 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__agentwatch__list_ai_crawlers"
],
"ask": [
"mcp__agentwatch__check_site",
"mcp__agentwatch__verify_crawler_visit",
"mcp__agentwatch__verify_signed_request"
],
"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 agentwatch. Review it before you use it.
Capability tags behind 2 rules
- mcp__agentwatch__check_site private-data, untrusted-input, outbound
- mcp__agentwatch__verify_signed_request 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/lighting.comiza/agentwatch/verify?hash=d4820e73fca7d623fef8ca6d483448c54afff3be8a1639d62a440db5a253b9b8" | jq -e '.changed == false'Pinned hash d4820e73fca7, from Nov 18, 2026, 17:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/lighting.comiza/agentwatch/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/lighting.comiza/agentwatch)Source: MCP Registry. Synced Oct 21, 2026, 18:53 UTC.
Agentwatch MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Agentwatch MCP server safe?
- No scan can promise that. What its definitions show: 4 tools, 2 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 Agentwatch MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the Agentwatch MCP server do things that cannot be undone?
- No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.
What tools does the Agentwatch MCP server have?
- 4 tools: 2 write, 2 read. verify_crawler_visit, verify_signed_request, check_site and list_ai_crawlers.
Does the Agentwatch MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the Agentwatch 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): check_site and verify_signed_request.
Has the Agentwatch MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 18, 2026.