hands.txt: a human proxy for AI agents MCP server
com.handstxt/human-proxyhandstxt.comstreamable-httpv1.0.0Scanned
Ask a real person to act in the physical world: a phone call, a visit, a fact checked on site.
11 tools: 1 can delete or cancel things, 5 do something that can't be undone, 3 change things and 2 only read.
- Reads private data · 1
- Reads text others write · 6
- Can send data out · 1
- 1 finding
Tools listed without sign-in. Remote posture
With Anlyon
1 tool here can delete or cancel things and 5 do something that can't be undone. 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 hands.txt: a human proxy for AI agents MCP server can do
The hands.txt: a human proxy for AI agents MCP server is published by handstxt.com, a domain the MCP Registry verified. It has 11 tools. It can delete or cancel things (1 tool: cancel_request), do things that cannot be undone, such as sending, paying or publishing (send_message, payment_address, check_payment, usdc_payment_details and 1 more) and create or change data (3 tools: start_test, submit_answers and create_request). 2 tools only read.
1 tool reads private data (get_request); 6 tools bring in text other people can write (create_request, get_request, send_message and 3 more); 1 tool can send data out of your system (send_message). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
The fixed checks matched 1 finding (annotation-conflict). Each one comes with the text it matched.
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.
11 tools
Deletes or cancels 1Claude Code: deny
cancel_requestCancels an open request.Reads text others writeNo findings
Why it is in this group
- The name "cancel_request" says "cancel": a word for removing or ending something.
- The description says "cannot be undone".
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Cancels an open request. This cannot be undone.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
Can't be undone 5Claude Code: ask
check_paymentOptional.Reads text others writeNo findings
Why it is in this group
- The name "check_payment" says "payment": a word that usually means money moves, which cannot be taken back.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Optional. Looks for your Bitcoin payment after its first confirmation and marks the request as paid. Pass txid only when payment_address said the address is shared by all requests.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
check_usdc_paymentOptional.Reads text others writeNo findings
Why it is in this group
- The name "check_usdc_payment" says "payment": a word that usually means money moves, which cannot be taken back.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Optional. Checks your USDC transfer once it has enough confirmations and marks the request as paid. Name the network you paid on and the transaction hash.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
payment_addressOptional.No findings
Why it is in this group
- The name "payment_address" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Optional. Returns the Bitcoin address for priority review of your request and the minimum amount. The payment covers priority review only; the service is free, the human may still decline, and there are no refunds.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
send_messageAdds a message to your request, for example an answer to the human's question.Reads text others writeCan send data outNo findings
Why it is in this group
- The name "send_message" says "send": a word for sending, and what is sent cannot be taken back.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe name "send_message" says "send": it sends or publishes something to other people.
Description, as the server wrote it Adds a message to your request, for example an answer to the human's question. You may write a few messages in a row; after that wait for the human to reply.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
usdc_payment_detailsOptional.1 finding
Why it is in this group
- The name "usdc_payment_details" says "payment": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "usdc_payment_details" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"openWorldHint":false}; The name "usdc_payment_details" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Optional. Returns the address, the networks and the exact USDC amount for priority review of your request. The amount is unique to the request: its last digits name it. The payment covers priority review only; the service is free, the human may still decline, and there are no refunds.
Annotations the server declares
- readOnlyHint: true
- openWorldHint: false
Changes things 3Claude Code: ask
create_requestFiles your request.Reads text others writeNo findings
Why it is in this group
- The name "create_request" says "create": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Files your request. A person reads it exactly as you write it, so be concrete: what to do, where, and what you want back. Do not include secrets. Returns the private URL of the request; keep it.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
start_testIssues the reverse CAPTCHA: a fictional log of about 3500 words with 12 questions.No findings
Why it is in this group
- The name "start_test" says "start": a word for a change.
Description, as the server wrote it Issues the reverse CAPTCHA: a fictional log of about 3500 words with 12 questions. The time limit starts when this tool answers, and each test allows one attempt. Returns the text and a challenge_id for submit_answers.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
submit_answersChecks your answers to the test.No findings
Why it is in this group
- The name "submit_answers" says "submit": a word for a change.
Description, as the server wrote it Checks your answers to the test. On success returns a pass for create_request, valid for 15 minutes and one request.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
Only reads 2Show 2 read-only tools
aboutReturns hands.txt: what the site does, its addresses, limits, rules and payment settings.No findings
Why it is in this group
- The description's first verb is "return": a word for reading.
Description, as the server wrote it Returns hands.txt: what the site does, its addresses, limits, rules and payment settings.
Annotations the server declares
- readOnlyHint: true
- openWorldHint: false
get_requestReturns the status of your request, the thread with the human and the payment state.Reads private dataReads text others writeNo findings
Why it is in this group
- The name "get_request" says "get": a word for reading.
What it can reach
- private-dataThe description's first sentence says "return" and "thread": it reads messages.
- untrusted-inputThe description's first sentence says "return" and "thread": it reads messages, which other people can write.
Description, as the server wrote it Returns the status of your request, the thread with the human and the payment state.
Annotations the server declares
- readOnlyHint: true
- openWorldHint: false
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.
6 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 outbound tool breaks every one.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write6
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out1
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
{
"permissions": {
"deny": [
"mcp__human-proxy__send_message"
]
}
}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, 15:18 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
- 10
- 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 10 tools that change things (2 destructive, 5 irreversible, 3 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://handstxt.com/mcp: tools/list returned 13 tools to an anonymous client, including cancel_request (destructive), erase_request (destructive), send_message (irreversible), payment_address (irreversible), check_payment (irreversible) and 5 more
Findings
Matches of fixed checks, with the text they matched.
The server marks "usdc_payment_details" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"openWorldHint":false}; The name "usdc_payment_details" says "payment": a word that usually means money moves, which cannot be taken back.
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 human-proxy https://handstxt.com/mcpclaude mcp add --transport http human-proxy https://handstxt.com/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__human-proxy__about",
"mcp__human-proxy__get_request"
],
"ask": [
"mcp__human-proxy__start_test",
"mcp__human-proxy__submit_answers",
"mcp__human-proxy__create_request",
"mcp__human-proxy__send_message",
"mcp__human-proxy__payment_address",
"mcp__human-proxy__check_payment",
"mcp__human-proxy__usdc_payment_details",
"mcp__human-proxy__check_usdc_payment"
],
"deny": [
"mcp__human-proxy__cancel_request"
]
}
}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 human-proxy. Review it before you use it.
Capability tags behind 6 rules
- mcp__human-proxy__create_request untrusted-input
- mcp__human-proxy__get_request private-data, untrusted-input
- mcp__human-proxy__send_message untrusted-input, outbound
- mcp__human-proxy__cancel_request untrusted-input
- mcp__human-proxy__check_payment untrusted-input
- mcp__human-proxy__check_usdc_payment untrusted-input
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.handstxt/human-proxy/verify?hash=089cb2a08228dcc5e1a712b3c14a72f4d395f46b30e08c18ca1b9d7c38e27fbc" | jq -e '.changed == false'Pinned hash 089cb2a08228, from Nov 17, 2026, 12:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/com.handstxt/human-proxy/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/com.handstxt/human-proxy)Source: MCP Registry. Synced Nov 17, 2026, 12:18 UTC.
hands.txt: a human proxy for AI agents MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the hands.txt: a human proxy for AI agents MCP server safe?
- No scan can promise that. What its definitions show: 11 tools, 9 of which can change, send or delete something, and 1 finding from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the hands.txt: a human proxy for AI agents MCP server delete data?
- Yes. 1 tool can delete, cancel or revoke something: cancel_request.
Can the hands.txt: a human proxy for AI agents MCP server do things that cannot be undone?
- Yes. 5 tools can send, pay, publish or merge: send_message, payment_address, check_payment, usdc_payment_details and check_usdc_payment.
What tools does the hands.txt: a human proxy for AI agents MCP server have?
- 11 tools: 1 destructive, 5 irreversible, 3 write, 2 read. cancel_request, send_message, payment_address, check_payment, usdc_payment_details, check_usdc_payment, start_test, submit_answers and 3 more.
Does the hands.txt: a human proxy for AI agents MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the hands.txt: a human proxy for AI agents MCP server send data to outside services?
- 1 tool can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): send_message.
Has the hands.txt: a human proxy for AI agents MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 17, 2026.