Skip to content

Agent Account Index

by mycelium-being · remote · no sign-in to list tools · v2026-10-12 · published by GitHub user mycelium-being

Can an AI agent hold an account at a service? Verdict, owner of record, quoted terms clause.

Changes · 1
CompareRegistry entry

2 tools: 1 change things and 1 only read.

Scanned Oct 21, 2026

No check matched anything in the tool definitions. No finding is not a safety guarantee. How the checks work.

claude mcp add --transport http agent-account-index https://agent-account-index.vercel.app/mcp

Tools

Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.

2 tools

Changes things 1Claude Code: ask
  • can_an_agent_hold_an_accountCheck whether a non-human actor may hold an account at a service.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 whether a non-human actor may hold an account at a service. Call this BEFORE creating an account anywhere. Returns the verdict, who answers for the account's acts, the verbatim clause from the service's own terms, and what the signup door demanded on load (captcha, phone field, SSO only). The stored captcha value disagreed with a browser re-read on 16 of the 54 doors whose value speaks, and not in one direction: 10 understated it, 5 named a widget the page lacked, 1 changed type. Read those against their own denominators: the 34 doors storing none_seen were wrong 10 times and the 20 storing a vendor name 6 times, about the same rate either way, so a disagreement does not say which reading to lean on. Where signup_door.captcha_rendered is not null it is that re-read ([] means re-read and nothing found); a captcha added on submit is invisible to both reads. These six integers are typed here and live in captcha_recheck_counts at the top of /index.json, which is the copy to trust. A row may also carry answered_by_a_named_human_2026_10_06: what a named person at the vendor answered when asked, held by the index with no page anyone can open. It sits beside the verdict and never changes it; the return counts the rows that carry one as rows_with_an_answered_reply. Null means nobody there was asked, which is not a finding.
Only reads 1Claude Code: allow
  • list_named_non_human_routesList every service that documents an identity a non-human actor may hold.Reads private dataNo findings

    Why it is in this group

    • The name "list_named_non_human_routes" says "list": a word for reading.

    What it can reach

    • private-dataThe description's first sentence says "list" and "document": it reads documents.
    Description, as the server wrote it
    List every service that documents an identity a non-human actor may hold. In 11 of the 12 a human or a legal entity remains owner of record and answers for the account's acts: there an agent acts as a delegate, not as the principal. 1 names no party at all (Wellknown): the holder is whoever proves control of a web origin and no identity is verified. Each row carries agent_as_principal.

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 untrusted-input or outbound.

Reads private data1
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 out0
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.

Install

Pinned to version 2026-10-12. Values in angle brackets are yours to fill in.

claude mcp add --transport http agent-account-index https://agent-account-index.vercel.app/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__agent-account-index__list_named_non_human_routes"
    ],
    "ask": [
      "mcp__agent-account-index__can_an_agent_hold_an_account"
    ],
    "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 agent-account-index. Review it before you use it.

Capability tags behind 1 rule
  • mcp__agent-account-index__list_named_non_human_routes private-data