Ned Watch MCP server
watch.ned/ned-watchned.watchstreamable-httpv1.2.0Scanned
Wake me when my agent misses its heartbeat. Deadman, overrun, HTTP, TLS, content; US+EU nodes.
12 tools: 1 can delete or cancel things, 10 change things and 1 only read.
- Reads private data · 0
- Reads text others write · 0
- Can send data out · 4
- No findings
Sign-in required, no OAuth metadata. Remote posture
Anlyon alternative
11 of these tools change something. Run those as governed actions.
Declare what an action changes, and Anlyon puts it under a shared limit, binds every approval to the exact request, and grades the outcome. It works on any HTTPS API you can describe as an action.
Anlyon Radar does not run, proxy or govern this MCP server. A tool your own code calls directly is not routed through Anlyon and is not governed by it.
Import an OpenAPI spec →What the Ned Watch MCP server can do
The Ned Watch MCP server is published by ned.watch, a domain the MCP Registry verified. It has 12 tools. It can delete or cancel things (1 tool: watch_cancel) and create or change data (10 tools: is_up, watch_register, quickstart, deadman_checkin and 6 more). 1 tool only reads.
4 tools can send data out of your system (watch_register, quickstart, overrun_watch and 1 more).
None of the fixed checks matched anything in its tool definitions. That is not a safety guarantee.
Last scanned Nov 5, 2026, version 1.2.0.
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.
12 tools
Deletes or cancels 1Claude Code: deny
watch_cancelCancel a watch you registered.No findings
Why it is in this group
- The name "watch_cancel" says "cancel": a word for removing or ending something.
Description, as the server wrote it Cancel a watch you registered. Frees a free-tier slot.
Changes things 10Claude Code: ask
balanceYour balance in cents, free allowance used, burn per day, days left, and the top-up routes.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 Your balance in cents, free allowance used, burn per day, days left, and the top-up routes.
content_watchFire when a URL answers but says the wrong thing.Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Fire when a URL answers but says the wrong thing. expect has 1 to 5 conditions: status (int), contains / not_contains (a string or list of strings), json_path (dotted, e.g. "data.items.0.state") with equals (a JSON scalar) or exists (true/false). No regex. Pages over 256 KB fail as 'size'. The callback names the condition that failed, never the page's content. 2c/day at >= 60s, 7c at >= 30s; free within the first 5 at >= 300s. Example: expect={"status": 200, "json_path": "status", "equals": "ok"}
deadman_checkinCheck in on a deadman watch to keep it from firing.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 Check in on a deadman watch to keep it from firing. Proves you hold the watch's signing_secret (returned at registration): it is sent as Authorization: Bearer <signing_secret> for this one call. No agent key needed.
is_upAsk before you call.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 before you call. No key needed. Returns the state of everything Ned watches publicly on a host (and its subdomains) from two continents: up, degraded (answered, but their own status page reports an incident), down, reachable (answers 401, handshake measured only) or unknown (not on The Watch). Also since when, the last change seen (tool list, version, auth) and how soon to ask again. Try api.openai.com, api.anthropic.com, mcp.context7.com, status.github.com. If unknown, the answer says how to add it (first five watches free).
overrun_finishThe run ended.No findings
Why it is in this group
- The description's first verb is "run": a word for a change.
Description, as the server wrote it The run ended. Returns runtime_s and late. If Ned already fired for this run, your callback gets 'clear' with late: true. Finishing with no run open is an error (409). failed=true says the run crashed: Ned fires at once with your short error (up to 200 chars) and clears when a later run finishes ok.
overrun_startA run began on an overrun watch.No findings
Why it is in this group
- The name "overrun_start" says "start": a word for a change.
Description, as the server wrote it A run began on an overrun watch. Returns run_id and deadline. Sends the watch's signing_secret as Authorization: Bearer for this one call; no agent key needed. run_id is optional (your own id, 1-64 chars); pass one and a retried start is safe (it returns the open run instead of superseding it).
overrun_watchFire when something is STILL running past its deadline (the other side of a deadman).Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Fire when something is STILL running past its deadline (the other side of a deadman). Register once, then call overrun_start when a run begins and overrun_finish when it ends. If a run isn't finished within max_runtime_s (60 to 86400), Ned POSTs one signed 'fire' (run_id, started_at, deadline); a finish after that sends 'clear' with late: true and the runtime. A start while a run is open closes the old one as superseded (no callback). label: give each job its own watch (the same registration twice returns the same watch). 1c/day; counts toward the free 5.
pricingNed Watch pricing: 5 free watches at >=300s, then per-day rates (http/tls/content 2c, fast 7c, deadman 1c, overrun 1c) from a prepaid balance topped up over x402 (USDC on Base).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 Ned Watch pricing: 5 free watches at >=300s, then per-day rates (http/tls/content 2c, fast 7c, deadman 1c, overrun 1c) from a prepaid balance topped up over x402 (USDC on Base). Public, no key needed. Paying is done against POST /v1/topup/{5,20,50}, not here.
quickstartThe fastest way to be watched, in one call.Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it The fastest way to be watched, in one call. Makes you an agent (if you have no key yet) and a deadman watch that expects a check-in every `every_minutes` (5 to 1440). Miss one and Ned POSTs a signed message to callback_url; check in again and he says it's clear. Returns agent_key (shown once), signing_secret, and the exact check-in line to run on your schedule. The first 5 watches are free. arm=true starts the clock now instead of at your first check-in, so a job that never runs at all still wakes you.
watch_registerRegister a watch.Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Register a watch. type: deadman | http | tls (for overrun and content use overrun_watch / content_watch). notify: up to 3 Discord/Slack/Teams webhook URLs told on fire, clear and change, so a human hears even if the agent behind callback_url is the thing that died. Your callback should outlive your agent. target: URL (http/tls); omit for deadman. interval_s >= 300 on the free tier. condition: e.g. {"grace_s": 600} for deadman, {"max_ms": 2000} for http latency, {"warn_days": 14} for tls. Ned POSTs a signed test callback to callback_url within 60s, then fires on change. No agent key yet? Call this without one: the response includes agent_key (shown once). Store it and pass it as NED_AGENT_KEY next time.
Only reads 1Show 1 read-only tool
watch_getCurrent state of one of your watches: status, fired, fire_count, next_check, last_state.No findings
Why it is in this group
- The name "watch_get" says "get": a word for reading.
Description, as the server wrote it Current state of one of your watches: status, fired, fire_count, next_check, last_state.
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.
No private-data, untrusted-input, outbound path in this server alone: no tool is tagged private-data or untrusted-input.
- Reads private data0
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write0
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out4
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
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, 19:21 UTC, unchanged since Nov 13, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 410
- tools/list served without credentials
- no
- Tools that change things, listed without credentials
- 0
- Protocol version
- not observed
- 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
The probe stopped: initialize answered HTTP 410.
Posture changes
- Nov 13, 2026
- HTTP status of an anonymous initialize: 200 to 410
- tools/list served without credentials: yes to no
- Tools that change things, listed without credentials: 11 to 0
- Protocol version: 2025-06-18 to none
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.
PostureRemote posture changed. HTTP status of an anonymous initialize: 200 -> 410; tools/list served without credentials: yes -> no; Tools that change things, listed without credentials: 11 -> 0 (and 1 more).View diff
- HTTP status of an anonymous initialize: 200 to 410
- tools/list served without credentials: yes to no
- Tools that change things, listed without credentials: 11 to 0
- Protocol version: 2025-06-18 to none
Install
Pinned to version 1.2.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http ned-watch https://api.ned.watch/mcpclaude mcp add --transport http ned-watch https://api.ned.watch/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__ned-watch__watch_get"
],
"ask": [
"mcp__ned-watch__is_up",
"mcp__ned-watch__watch_register",
"mcp__ned-watch__quickstart",
"mcp__ned-watch__deadman_checkin",
"mcp__ned-watch__overrun_watch",
"mcp__ned-watch__overrun_start",
"mcp__ned-watch__overrun_finish",
"mcp__ned-watch__content_watch",
"mcp__ned-watch__balance",
"mcp__ned-watch__pricing"
],
"deny": [
"mcp__ned-watch__watch_cancel"
]
}
}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 ned-watch. Review it before you use it.
Capability tags behind 4 rules
- mcp__ned-watch__watch_register outbound
- mcp__ned-watch__quickstart outbound
- mcp__ned-watch__overrun_watch outbound
- mcp__ned-watch__content_watch 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/watch.ned/ned-watch/verify?hash=47bc84f13b2747fc48d068b047e6bbd28a285f4f14085dfc626f942f54a31cfe" | jq -e '.changed == false'Pinned hash 47bc84f13b27, from Nov 5, 2026, 15:17 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/watch.ned/ned-watch/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/watch.ned/ned-watch)Source: MCP Registry. Synced Oct 21, 2026, 18:55 UTC.
Ned Watch MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Ned Watch MCP server safe?
- No scan can promise that. What its definitions show: 12 tools, 11 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 Ned Watch MCP server delete data?
- Yes. 1 tool can delete, cancel or revoke something: watch_cancel.
Can the Ned Watch 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 Ned Watch MCP server have?
- 12 tools: 1 destructive, 10 write, 1 read. watch_cancel, is_up, watch_register, quickstart, deadman_checkin, overrun_watch, overrun_start, overrun_finish and 4 more.
Does the Ned Watch MCP server need sign-in or OAuth?
- Its remote endpoint did not answer the last anonymous look.
Does the Ned Watch MCP server send data to outside services?
- 4 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): watch_register, quickstart, overrun_watch and content_watch.
Has the Ned Watch MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 5, 2026.