Skip to content

Relaystation MCP server

ai.relaystation/relaystationrelaystation.aistreamable-httpv1.0.0Scanned

Pay-per-call agent infrastructure: file storage & handoff, compute tools, messaging, e-sign, KYC.

14 tools: 2 do something that can't be undone, 9 change things and 3 only read.

Tools: 0 delete, 2 irreversible, 9 write, 3 read.
  • Reads private data · 0
  • Reads text others write · 0
  • Can send data out · 2
  • 3 findings

Tools listed without sign-in. Remote posture

With Anlyon

2 tools here 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.
Govern this server →

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 Relaystation MCP server can do

The Relaystation MCP server is published by relaystation.ai, a domain the MCP Registry verified. It has 14 tools. It can do things that cannot be undone, such as sending, paying or publishing (ask_operator and message_agent) and create or change data (9 tools: create_agent_address, create_baton, append_to_baton, mint_token and 5 more), but cannot delete data. 3 tools only read.

2 tools can send data out of your system (ask_operator and message_agent).

The fixed checks matched 3 findings (credentials-mention). Each one comes with the text it matched.

Its remote endpoint lists its tools without sign-in.

Last scanned Nov 11, 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.

14 tools

Can't be undone 2Claude Code: ask
  • ask_operatorSend a question to the human operator via their Telegram and wait for their reply.Can send data outNo findings

    Why it is in this group

    • The description's first verb is "send": a word for sending, and what is sent cannot be taken back.

    What it can reach

    • outboundThe description's first verb is "send": it sends or publishes something to other people.
    Description, as the server wrote it
    Send a question to the human operator via their Telegram and wait for their reply. Use when you need a decision, approval, or clarification you can't infer from the conversation. The operator gets a Telegram notification on their phone; their reply comes back to you here. Typical reply latency: 1-15 minutes when operator is away from desk; ~10 seconds when at-desk. Cost: $0.005 above the daily free tier (5 free/day). If the operator has disabled the bridge ("at desk"), this returns immediately with a 'use in-IDE interaction' hint — fall through to the host IDE's chat surface. For long messages (>4KB), use ask_operator_email (needs an OAuth-linked email address — wallet-only customers should attach a Baton instead) or attach a Baton via attached_baton_id (saves the long content as a re-readable workspace; $0.01). ASK = blocks until the operator replies (vs notify_operator = one-way, no reply). Example: ask_operator {message:'Deploy to prod now?'}
  • message_agentSend a message to another Relaystation agent's ephemeral address (format <local_part>@courier.relaystation.ai).Can send data outNo findings

    Why it is in this group

    • The description's first verb is "send": a word for sending, and what is sent cannot be taken back.

    What it can reach

    • outboundThe description's first verb is "send": it sends or publishes something to other people.
    Description, as the server wrote it
    Send a message to another Relaystation agent's ephemeral address (format <local_part>@courier.relaystation.ai). Same-Relaystation routing only (v2 launch); external domains return EXTERNAL_DOMAIN_NOT_SUPPORTED_V2. Free — agent-to-agent traffic has no per-message cost (subject to the recipient's per-address inbox quota; default 100 messages within the 30-day retention window). Use for multi-agent coordination: delegating subtasks, sharing context, building chains. For ongoing coordination involving 30+ messages, prefer a shared LEDGER or SCRATCHPAD Baton (better search/audit/verification at $0.01 setup). Example: message_agent {to_address:'agent-7@courier.relaystation.ai', body:'Status update: task ready'}
Changes things 9Claude Code: ask
  • append_to_batonWrite to a baton — append (for append / append-chained presets) or overwrite (for single-object / checkpoint presets).No findings

    Why it is in this group

    • The description's first verb is "write": a word for a change.
    Description, as the server wrote it
    Write to a baton — append (for append / append-chained presets) or overwrite (for single-object / checkpoint presets). FREE (consumes the prepaid writes budget; no per-call charge). Owner or editor/write-token only. Set overwrite=true to replace in full on overwrite-shaped batons. Example: append_to_baton {id:"bat_<id>", content:"<text or base64>"}
  • balanceYour current Relaystation balance and available (post-hold) credit, plus account profile.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 current Relaystation balance and available (post-hold) credit, plus account profile. Read-only, FREE. Requires your rs_live_* key (or wallet-JWT / MCP token); scoped to YOUR account only. Example: balance {}
  • call_toolRun a tool found via search_tools by name, passing its arguments.1 finding

    Why it is in this group

    • The name "call_tool" says "call": a word for a change.

    Findings

    • Info

      The description mentions "credential".

      Matched text

      description: "...l. Consequential tools (account, money, credential, messaging — e.g. create_topup_link, cr..."

      What this check looks for
    Description, as the server wrote it
    Run a tool found via search_tools by name, passing its arguments. Dispatches the safe/billable utility tools (the cputools catalog + free reads/quotes) through the identical auth + billing + validation as a direct call. Consequential tools (account, money, credential, messaging — e.g. create_topup_link, create_baton, mint_token, ask_operator) are NOT dispatchable here: call them directly by name so your client can gate them with per-tool consent. FREE to invoke (the wrapped tool bills itself). Example: call_tool {name:"pdf_merge", arguments:{files:[{inputKey:"<key1>"},{inputKey:"<key2>"}]}}
  • create_agent_addressMint a new ephemeral Relaystation address for this agent to receive messages from other agents.No findings

    Why it is in this group

    • The name "create_agent_address" says "create": a word for a change.
    Description, as the server wrote it
    Mint a new ephemeral Relaystation address for this agent to receive messages from other agents. Returns a unique <local_part>@courier.relaystation.ai address tied to this agent's customer. Free up to the daily mint cap (default 10/day per customer); above-cap mints debit bridge.agent_address.over_cap_price_micros (default $0.001) via Pattern A. purpose_label is informational. ttl_seconds defaults to 24h; addresses past TTL bounce inbound messages. Example: create_agent_address {purpose_label:"inbox-for-task-42"}
  • create_batonCreate a baton — a prepaid storage object.No findings

    Why it is in this group

    • The name "create_baton" says "create": a word for a change.
    Description, as the server wrote it
    Create a baton — a prepaid storage object. Pick a preset (drop=store a file, pass=share a file, scratchpad=collaborate via append log, checkpoint=a single current document that each write REPLACES, ledger=hash-chained tamper-evident log) and optional tier, or supply a customShape. **To collaborate on a document — human↔human, human↔agent or agent↔agent — use `checkpoint`**: it holds one current state, so every reader sees the same document and a write replaces it rather than piling up entries. `scratchpad` is an append log: good for a running buffer two agents both add to, but readers of a scratchpad see the newest entry only. BILLABLE: charged once at create (the engine quote of the shape; quote it first with quote_baton). Funded from your balance via your rs_live_* key. Pass inline `content` (≤3MB; base64 for binary) or omit and write later with append_to_baton. flags.hashChaining=true makes an append log tamper-evident (auto-includes a prepaid witness). NOTE: hash-chained batons (ledger preset, or flags.hashChaining=true) MUST be created empty — do NOT pass `content` (returns 422 CHAINED_BATON_REQUIRES_EMPTY_CREATE); create empty, then append_to_baton for entry 1. Example (store): create_baton {preset:"drop", tier:"femto", content:"<base64>"}. Example (hash-chained): create_baton {preset:"ledger", tier:"femto", flags:{hashChaining:true}} then append_to_baton.
  • create_contractCreate a binding-by-goodwill agreement between agents/parties and get back a contractId + termsHash + per-signer read tokens.No findings

    Why it is in this group

    • The name "create_contract" says "create": a word for a change.
    Description, as the server wrote it
    Create a binding-by-goodwill agreement between agents/parties and get back a contractId + termsHash + per-signer read tokens. The contract IS a hash-chained, auto-witnessed record (terms = entry 1, consents append after); it is tamper-evident and permanently verifiable, but it is NOT legally binding, NOT court-enforceable, and involves NO escrow or money movement — enforcement rests on the good faith of the parties. BILLABLE (contracts.create, charge-on-attempt). The system appends a canonical signature block (and, if arbitration:true, an arbitration block) to your terms — your terms body must NOT itself contain those blocks. Each required signer is identified by EXACTLY one method: a "wallet" (0x-address, satisfied only by an EIP-712 ContractConsent signature) or an "account" (a Relaystation customerId, satisfied only by that authenticated account). Input: { terms (required), requiredSigners: [{ identity, kind: "wallet"|"account" }] (required, ≥1), arbitration?: boolean, signingWindowSeconds?: int (default 72h) }. Requires an Idempotency-Key. Returns { id, batonId, termsHash, status, signingDeadline, requiredSigners, tokens: [{ identity, tokenId }] }. Hand each signer their read token + the contractId + termsHash so they can review and sign. Example: create_contract {terms:"Both parties agree…", requiredSigners:[{identity:"0xABC…", kind:"wallet"}, {identity:"<customerId>", kind:"account"}]}.
  • mint_tokenMint a collaborator token for a baton so another agent — or a human — can read and/or write without your key.1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.

    Findings

    • Info

      The description mentions "credential".

      Matched text

      description: "...token id (tok_…) to share as the baton credential, plus an `editUrl` a human can open in..."

      What this check looks for
    Description, as the server wrote it
    Mint a collaborator token for a baton so another agent — or a human — can read and/or write without your key. FREE. type is read / write / read_write; optionally cap reads_allowed / writes_allowed, set an expiry, or restrict by IP. Up to 100 tokens per baton. Owner only. Returns the token id (tok_…) to share as the baton credential, plus an `editUrl` a human can open in a browser to read and edit the content directly. **Works on any baton; best on a `checkpoint`** — that is the document shape, where a save replaces the current state so every reader (human or agent) sees the same thing. On append-shaped batons (scratchpad, ledger) a save adds an entry and the editor shows the newest entry only. Example: mint_token {id:"bat_<id>", type:"read_write"}
  • sign_contractAppend your consent to a contract (one entry on its hash chain).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
    Append your consent to a contract (one entry on its hash chain). FREE. Two forms, no cross-method satisfaction: a "wallet" signer supplies an EIP-712 ContractConsent signature over the contract's termsHash (the signature IS the authentication — no account needed); an "account" signer must be authenticated as the matching Relaystation customerId. When the LAST required signer signs, the contract becomes "executed" and is witnessed. The EIP-712 typed data is domain { name:"Relaystation Contracts", version:"1", chainId } / ContractConsent(string contractId, bytes32 termsHash). Input: { id (contractId), method:"wallet", walletAddress, signature } OR { id, method:"account" }. Returns the updated contract view. Example: sign_contract {id:"<contract-id>", method:"account"}
Only reads 3Show 3 read-only tools
  • describe_toolGet one tool's full description and input schema by exact name (as returned by search_tools).No findings

    Why it is in this group

    • The name "describe_tool" says "describe": a word for reading.
    Description, as the server wrote it
    Get one tool's full description and input schema by exact name (as returned by search_tools). FREE. Example: describe_tool {name:'pdf_merge'}
  • read_batonRead a baton's content + metadata.1 finding

    Why it is in this group

    • The name "read_baton" says "read": a word for reading.

    Findings

    • Info

      The description mentions "credential".

      Matched text

      description: "...rator token id (tok_…, the token IS the credential). Large multi-entry batons return a pre..."

      What this check looks for
    Description, as the server wrote it
    Read a baton's content + metadata. FREE (reads draw down the prepaid egress budget, no per-call charge). Address by owner id (requires your key) or by a collaborator token id (tok_…, the token IS the credential). Large multi-entry batons return a presigned download URL or a hint to page via list-entries; small content returns inline. Example: read_baton {id:"bat_<id>"}
  • search_toolsSearch the full Relaystation tool catalog by keyword and get back the best matches.No findings

    Why it is in this group

    • The name "search_tools" says "search": a word for reading.
    Description, as the server wrote it
    Search the full Relaystation tool catalog by keyword and get back the best matches. Use this first to find a tool, then `describe_tool` for its schema and `call_tool` to run it (or call a named hot tool directly). `query` is free text (e.g. "merge pdf", "csv to json", "send telegram"). `detail` controls how much is returned per match: "name" | "summary" (default) | "full" (with inputSchema). `limit` defaults to 5. FREE. Example: search_tools {query:"merge pdf", limit:5}

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 out2
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.

Tools listed without sign-in

Looked at Nov 20, 2026, 18:27 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
11
Protocol version
2025-06-18
WWW-Authenticate on 401
not observed
resource_metadata in WWW-Authenticate
not observed
RFC 9728 protected resource metadata
https://api.relaystation.ai/.well-known/oauth-protected-resource/mcp
Authorization servers
https://api.relaystation.ai
Legacy HTTP+SSE transport
no
Dynamic client registration
yes
Client ID metadata documents
no
  • HighThe whole server

    The endpoint served tools/list without credentials, and the list has 11 tools that change things (2 irreversible, 9 write). Whether a call also works without credentials was not tested: tools are never called.

    Matched text

    https://api.relaystation.ai/mcp: tools/list returned 14 tools to an anonymous client, including ask_operator (irreversible), message_agent (irreversible), create_agent_address (write), create_baton (write), append_to_baton (write) and 6 more

    What this check looks for

Findings

Matches of fixed checks, with the text they matched.

3 info
  • The description mentions "credential".

    Matched text

    description: "...rator token id (tok_…, the token IS the credential). Large multi-entry batons return a pre..."

    What this check looks for
  • The description mentions "credential".

    Matched text

    description: "...token id (tok_…) to share as the baton credential, plus an `editUrl` a human can open in..."

    What this check looks for
  • The description mentions "credential".

    Matched text

    description: "...l. Consequential tools (account, money, credential, messaging — e.g. create_topup_link, cr..."

    What this check looks for

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 relaystation https://api.relaystation.ai/mcp
claude mcp add --transport http relaystation https://api.relaystation.ai/mcp

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.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__relaystation__read_baton",
      "mcp__relaystation__search_tools",
      "mcp__relaystation__describe_tool"
    ],
    "ask": [
      "mcp__relaystation__ask_operator",
      "mcp__relaystation__message_agent",
      "mcp__relaystation__create_agent_address",
      "mcp__relaystation__create_baton",
      "mcp__relaystation__append_to_baton",
      "mcp__relaystation__mint_token",
      "mcp__relaystation__balance",
      "mcp__relaystation__create_topup_link",
      "mcp__relaystation__call_tool",
      "mcp__relaystation__create_contract",
      "mcp__relaystation__sign_contract"
    ],
    "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 relaystation. Review it before you use it.

Capability tags behind 2 rules
  • mcp__relaystation__ask_operator outbound
  • mcp__relaystation__message_agent 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/ai.relaystation/relaystation/verify?hash=57728a78134439f183454d54c9e4973fcb62d82e32360a57fc4e86bfa740b0c7" | jq -e '.changed == false'

Pinned hash 57728a781344, from Nov 11, 2026, 21:17 UTC.

Watch

Get told when its tools change: a new tool, a new description, a new class.

https://radar.anlyon.com/servers/ai.relaystation/relaystation/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Badge

Anlyon Radar badge for ai.relaystation/relaystation
[![Anlyon Radar](https://radar.anlyon.com/badge/ai.relaystation/relaystation.svg)](https://radar.anlyon.com/servers/ai.relaystation/relaystation)

Source: MCP Registry. Synced Oct 21, 2026, 17:43 UTC.

Relaystation MCP server: questions and answers

Answered from the scan above, not from the publisher's description.

Is the Relaystation MCP server safe?

No scan can promise that. What its definitions show: 14 tools, 11 of which can change, send or delete something, and 3 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 Relaystation MCP server delete data?

No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.

Can the Relaystation MCP server do things that cannot be undone?

Yes. 2 tools can send, pay, publish or merge: ask_operator and message_agent.

What tools does the Relaystation MCP server have?

14 tools: 2 irreversible, 9 write, 3 read. ask_operator, message_agent, create_agent_address, create_baton, append_to_baton, mint_token, balance, create_topup_link and 6 more.

Does the Relaystation MCP server need sign-in or OAuth?

Its remote endpoint listed its tools to an anonymous client, without sign-in.

Does the Relaystation 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): ask_operator and message_agent.

Has the Relaystation MCP server changed its tools?

Not since Anlyon Radar started tracking it. Last scanned Nov 11, 2026.