Skip to content

PubSpace MCP server

ai.pubspace/mcppubspace.aistreamable-httpv0.25.1Scanned

Publish HTML or markdown from your agent as a shareable link, with password or work-email walls.

12 tools: 2 can delete or cancel things, 1 do something that can't be undone, 6 change things and 3 only read.

Tools: 2 delete, 1 irreversible, 6 write, 3 read.
  • Reads private data · 4
  • Reads text others write · 10
  • Can send data out · 2
  • 3 findings, 1 high

Tools listed without sign-in. Remote posture

With Anlyon

2 tools here can delete or cancel things and 1 does 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 PubSpace MCP server can do

The PubSpace MCP server is published by pubspace.ai, a domain the MCP Registry verified. It has 12 tools. It can delete or cancel things (2 tools: update_page and delete_page), do things that cannot be undone, such as sending, paying or publishing (publish_page) and create or change data (6 tools: unpublish_page, resolve_comment, set_comments, create_account and 2 more). 3 tools only read.

4 tools read private data (list_pages, get_page, list_comments and 1 more); 10 tools bring in text other people can write (publish_page, update_page, list_pages and 7 more); 2 tools can send data out of your system (publish_page and set_comments). 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 3 findings, 1 of them high severity (sensitive-path and credentials-mention). Each one comes with the text it matched.

Its remote endpoint lists its tools without sign-in, and its tools changed once since Anlyon Radar first saw them.

Last scanned Nov 15, 2026, version 0.25.1.

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 2Claude Code: deny
  • delete_pageDelete a page: its URL stops working at once.Reads text others writeNo findings

    Why it is in this group

    • The name "delete_page" says "delete": a word for removing or ending something.
    • 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
    Delete a page: its URL stops working at once. The maker gets only a brief undo in the dashboard, then it is purged for good (there is no restore tool). Only use it when the human asked for deletion; to hide a page reversibly, use unpublish_page instead. Returns { ok, id, deleted }; fails with page_not_found if the id is not in this space or is already deleted.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: true
    • idempotentHint: false
    • openWorldHint: true
  • update_pageReplace the content of an existing page in place: the same URL serves the new content immediately, so links the human already shared keep working.Reads text others write1 finding

    Why it is in this group

    • The name "update_page" says "update": a word for a change.
    • 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.

    Findings

    • Info

      The description mentions "credentials".

      Matched text

      description: "...he new content looks like it holds live credentials. A page that works as a sign-in or paym..."

      What this check looks for
    Description, as the server wrote it
    Replace the content of an existing page in place: the same URL serves the new content immediately, so links the human already shared keep working. Use this instead of publish_page whenever the page already exists. The previous version stays restorable from the dashboard for 7 days (Pro keeps full version history). Pass draft: true to stage the update instead: nothing changes for viewers until the maker reviews the previewUrl and presses "Sync to live"; use it when the human asked to see changes before they go out. Fails with page_not_found if the id is not a live or offline page in this space, and with secrets_detected (nothing changes) if the new content looks like it holds live credentials. A page that works as a sign-in or payment form sending what people type to another site can be refused with unsafe_page_detected (no override): make such forms non-functional demos. Connected without an API key, it updates a preview instead: pass the claim_token that publish_page returned (or omit it in the same session that published the page). The URL stays the same and the 7-day expiry does not move; previews have no drafts or version history. After the human claims the preview (claim_page), a page that moved into their account answers preview_already_claimed: reconnect with their API key to keep updating it. Pages that did not fit their plan stay previews and the claim token still updates them.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: true
    • idempotentHint: true
    • openWorldHint: true
Can't be undone 1Claude Code: ask
  • publish_pagePublish a new HTML (or Markdown) page to the maker's PubSpace space and get back a live, shareable URL to hand to the human.Reads text others writeCan send data out2 findings

    Why it is in this group

    • The name "publish_page" says "publish": a word for a change that another call does not undo.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe name "publish_page" says "publish": it sends or publishes something to other people.

    Findings

    • High

      The description refers to ".env".

      Matched text

      description: "...ve credentials (API keys, private keys, .env lines, database URLs with passwords) is..."

      What this check looks for
    • Info

      The description mentions "credentials".

      Matched text

      description: "...the human. Content that looks like live credentials (API keys, private keys, .env lines, da..."

      What this check looks for
    Description, as the server wrote it
    Publish a new HTML (or Markdown) page to the maker's PubSpace space and get back a live, shareable URL to hand to the human. Pages are unlisted by default — indexed by search engines only if the maker turns that on later (Pro); anyone with the link can view. Use this only for a brand-new page; to change a page that already exists, call update_page so its URL stays the same (find the id with list_pages). Each call creates a page that counts toward the plan's live-page cap; at the cap the call fails with doc_limit_reached and, on the free plan, an upgrade link to relay to the human. Content that looks like live credentials (API keys, private keys, .env lines, database URLs with passwords) is refused with secrets_detected and the line numbers; remove them and retry. A page that works as a sign-in or payment form sending what people type to another site can be refused with unsafe_page_detected (no override): make such forms non-functional demos. Returns { url, id, mode, visibility, dashboardUrl }. Connected without an API key, it publishes a 7-day preview at a pubspace.page link instead (no account needed, up to 5 MB, no password, stricter sign-in lookalike check) and returns { url, id, mode, preview: true, expiresAt, claimToken, tell_user }: relay tell_user, and keep claimToken, the only way to change that page later (update_page with claim_token). Pass claim_token to add another page under the same token (up to 5 live previews). Later publishes in the same session join the first preview's token and do not return claimToken again: that first token covers them. To keep previews for good, call claim_page with the human's email.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
Changes things 6Claude Code: ask
  • claim_pageFor pages published without an API key (previews): ask PubSpace to email your human one link that keeps every page under a claim token.Reads text others writeNo 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

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    For pages published without an API key (previews): ask PubSpace to email your human one link that keeps every page under a claim token. Callable without an API key. Only call it after the human has told you their email; never invent or guess one. Opening the link shows a Keep button; pressing it, or one bare POST to the link (do that if the human pastes it to you), moves the pages into a free PubSpace space at <name>.pubspace.ai (or into the account they already have, as many pages as its plan holds): the preview links redirect there and stop expiring, even after the 7 days. Pass claim_token from publish_page, or omit it in the session that published the preview. Returns { ok, sent, email, expiresInHours, tell_user }: relay tell_user. The answer is the same whether or not the address already has an account. Once the human clicks, the pages that moved answer preview_already_claimed: ask them for their API key to keep updating those. Pages that did not fit an existing account's plan stay previews; the claim token still updates them, and calling claim_page again keeps them once there is room. When every page has moved, the claim token stops working. Fails with invalid_email, disposable_email_not_allowed, email_opted_out, invalid_claim_token, preview_already_claimed or rate_limited.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • create_accountSign your human up for a free PubSpace space (no card) without leaving this session.Reads text others writeNo findings

    Why it is in this group

    • The name "create_account" 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
    Sign your human up for a free PubSpace space (no card) without leaving this session. Callable without an API key. Only call it after the human has told you their email and chosen a space name; never invent or guess either. Their pages will live at <subdomain>.pubspace.ai, so treat the name as permanent (renames are limited). A 6-digit verification code is emailed to them: ask them to read it to you, then call verify_email. Once verified, publish_page works in this same session (if your client sends no Mcp-Session-Id, follow verifyHint instead). Returns { subdomain, email, apiKey, verificationRequired, connectHint }; persist the key with the connectHint command so future sessions start connected. Fails with email_taken (they already have an account: have them sign in and copy their key from Settings → Connect your agent), subdomain_taken, invalid_subdomain, invalid_email or rate_limited. If this session is already connected with a key, it fails with already_connected.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • resolve_commentMark a comment thread resolved (or reopen it with resolved: false) after you have addressed it.Reads private dataReads text others writeNo findings

    Why it is in this group

    • The name "resolve_comment" says "resolve": a word for reading.
    • The description's first verb is "mark": a word for a change.

    What it can reach

    • private-dataThe description's first sentence says "thread": it reads messages.
    • untrusted-inputThe name "resolve_comment" says "resolve" and "comment": it reads comments, which other people can write.
    • untrusted-inputThe description's first sentence says "mark" and "comment": it reads comments, which other people can write.
    Description, as the server wrote it
    Mark a comment thread resolved (or reopen it with resolved: false) after you have addressed it. The pin turns grey and drops out of the open list for everyone; nothing is deleted. Returns { ok, page_id, thread_id, resolved }. Fails with thread_not_found if the id is not an open-or-resolved thread on that page.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • set_commentsTurn click-to-pin comments on or off for a page.Can send data outNo findings

    Why it is in this group

    • The name "set_comments" says "set": a word for a change.

    What it can reach

    • outboundThe name "set_comments" says "set" and "comment": it writes something other people read.
    Description, as the server wrote it
    Turn click-to-pin comments on or off for a page. While on, anyone who can view the page can click a spot on it to comment once they sign in with their email (on a work-email-locked page, their viewing sign-in is enough); the maker gets an email per comment. Off hides comments without deleting them. Off by default. Returns { ok, page_id, commentsEnabled }. Use it when the human asks for feedback or review on a page.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • unpublish_pageTake a page offline without deleting it: its URL shows an "offline" notice to viewers, and the content, id, URL and view count are kept.Reads text others writeNo 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

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    Take a page offline without deleting it: its URL shows an "offline" notice to viewers, and the content, id, URL and view count are kept. Pass restore: true to put it back online at the same URL. Prefer this over delete_page when the human wants a page hidden or paused. Calling it twice is harmless. Returns { ok, id, live }; fails with page_not_found if the id is not in this space.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • verify_emailFinish account verification with the 6-digit code PubSpace emailed to the human (ask them to read it to you).No findings

    Why it is in this group

    • The input takes a statement or command to run (input.code), which can change anything it can reach.
    Description, as the server wrote it
    Finish account verification with the 6-digit code PubSpace emailed to the human (ask them to read it to you). Publishing unlocks on success. Works in the session where create_account ran, with an API key for an unverified account, or on any connection by passing api_key (the apiKey create_account returned); in that last case, reconnect with the key as the X-API-Key header to publish. Returns { ok, verified, subdomain }; fails with invalid_code (and attemptsRemaining), code_expired, too_many_attempts, already_verified, not_signed_in or invalid_api_key.

    Annotations the server declares

    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
Only reads 3Show 3 read-only tools
  • get_pageFetch one page by id: metadata always ({ id, title, url, project, views, live, workEmailGated, hasDraft, comments: { total, open }, createdAt, updatedAt, visibility }), plus the st…Reads private dataReads text others writeNo findings

    Why it is in this group

    • The name "get_page" says "get": a word for reading.

    What it can reach

    • private-dataThe description's first sentence says "fetch" and "email": it reads mail.
    • untrusted-inputThe description's first sentence says "fetch" and "url": it reads web pages, which anyone can write.
    Description, as the server wrote it
    Fetch one page by id: metadata always ({ id, title, url, project, views, live, workEmailGated, hasDraft, comments: { total, open }, createdAt, updatedAt, visibility }), plus the stored HTML when include_html is true. Read-only. Use it to read a page before editing it with update_page; use list_pages to find ids or browse. Fails with page_not_found if the id is not in this space or the page was deleted.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • list_commentsRead the comments people left on a page: { page_id, url, commentsEnabled, threads: [{ id, resolved, on, comments: [{ id, author, body, createdAt }] }] }.Reads private dataReads text others writeNo findings

    Why it is in this group

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

    What it can reach

    • private-dataThe description's first sentence says "read" and "thread": it reads messages.
    • untrusted-inputThe name "list_comments" says "list" and "comment": it reads comments, which other people can write.
    • untrusted-inputThe description's first sentence says "read" and "comment": it reads comments, which other people can write.
    Description, as the server wrote it
    Read the comments people left on a page: { page_id, url, commentsEnabled, threads: [{ id, resolved, on, comments: [{ id, author, body, createdAt }] }] }. Each thread is a pin someone dropped on the page; `on` is the text of the element they clicked, so you can tell what the comment is about without seeing the page. author is the commenter's full email. Read-only. Typical loop: list_comments → update_page with the fixes → resolve_comment on each thread you addressed. Open threads only unless include_resolved is true. Fails with page_not_found if the id is not in this space.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • list_pagesList every page in the maker's space, newest first, with no paging: { total, pages: [{ id, title, url, project, views, live, workEmailGated, hasDraft, comments: { total, open }, cr…Reads private dataReads text others writeNo findings

    Why it is in this group

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

    What it can reach

    • private-dataThe description's first sentence says "list" and "email": it reads mail.
    • untrusted-inputThe description's first sentence says "list" and "url": it reads web pages, which anyone can write.
    Description, as the server wrote it
    List every page in the maker's space, newest first, with no paging: { total, pages: [{ id, title, url, project, views, live, workEmailGated, hasDraft, comments: { total, open }, createdAt, updatedAt }] }. Read-only. Use it to find a page_id before update_page / get_page / unpublish_page / delete_page, or to reuse an existing project name. Offline pages are included (live: false); deleted pages are not. Pass q to search instead of scanning the full list; use get_page when you need one page's HTML.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: 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.

80 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 2 outbound tools breaks every one.

Reads private data4
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write10
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.
deny to break every path
{
  "permissions": {
    "deny": [
      "mcp__pubspace__publish_page",
      "mcp__pubspace__set_comments"
    ]
  }
}

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 17, 2026, 15:23 UTC, unchanged since Nov 10, 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
9
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 9 tools that change things (2 destructive, 1 irreversible, 6 write). Whether a call also works without credentials was not tested: tools are never called.

    Matched text

    https://pubspace.ai/mcp: tools/list returned 12 tools to an anonymous client, including update_page (destructive), delete_page (destructive), publish_page (irreversible), unpublish_page (write), resolve_comment (write) and 4 more

    What this check looks for

Posture changes

  1. Nov 10, 2026
    • Tools that change things, listed without credentials: 6 to 9
  2. Nov 3, 2026
    • Tools that change things, listed without credentials: 4 to 6

Findings

Matches of fixed checks, with the text they matched.

1 high2 info
  • The description refers to ".env".

    Matched text

    description: "...ve credentials (API keys, private keys, .env lines, database URLs with passwords) is..."

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

    Matched text

    description: "...the human. Content that looks like live credentials (API keys, private keys, .env lines, da..."

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

    Matched text

    description: "...he new content looks like it holds live credentials. A page that works as a sign-in or paym..."

    What this check looks for

History

Changes to its tool definitions and to its remote endpoint, newest first.

  1. Tools4 added, 6 changed.+4 added~6 changedView diff
  2. PostureRemote posture changed. Tools that change things, listed without credentials: 6 -> 9.View diff
    • Tools that change things, listed without credentials: 6 to 9
  3. PostureRemote posture changed. Tools that change things, listed without credentials: 4 -> 6.View diff
    • Tools that change things, listed without credentials: 4 to 6

Install

Pinned to version 0.25.1. Values in angle brackets are yours to fill in.

claude mcp add --transport http pubspace https://pubspace.ai/mcp
claude mcp add --transport http pubspace https://pubspace.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__pubspace__list_pages",
      "mcp__pubspace__get_page",
      "mcp__pubspace__list_comments"
    ],
    "ask": [
      "mcp__pubspace__publish_page",
      "mcp__pubspace__unpublish_page",
      "mcp__pubspace__resolve_comment",
      "mcp__pubspace__set_comments",
      "mcp__pubspace__create_account",
      "mcp__pubspace__verify_email",
      "mcp__pubspace__claim_page"
    ],
    "deny": [
      "mcp__pubspace__update_page",
      "mcp__pubspace__delete_page"
    ]
  }
}

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 pubspace. Review it before you use it.

Capability tags behind 11 rules
  • mcp__pubspace__publish_page untrusted-input, outbound
  • mcp__pubspace__update_page untrusted-input
  • mcp__pubspace__list_pages private-data, untrusted-input
  • mcp__pubspace__get_page private-data, untrusted-input
  • mcp__pubspace__unpublish_page untrusted-input
  • mcp__pubspace__delete_page untrusted-input
  • mcp__pubspace__list_comments private-data, untrusted-input
  • mcp__pubspace__resolve_comment private-data, untrusted-input
  • mcp__pubspace__set_comments outbound
  • mcp__pubspace__create_account untrusted-input
  • mcp__pubspace__claim_page 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/ai.pubspace/mcp/verify?hash=4483232211e4066101089a158cd1f310a8dc28b68d810e00fd31663ec526ad6b" | jq -e '.changed == false'

Pinned hash 4483232211e4, from Nov 15, 2026, 02:17 UTC.

Watch

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

https://radar.anlyon.com/servers/ai.pubspace/mcp/changes.xml

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

Badge

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

Source: MCP Registry. Synced Nov 2, 2026, 01:17 UTC.

PubSpace MCP server: questions and answers

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

Is the PubSpace MCP server safe?

No scan can promise that. What its definitions show: 12 tools, 9 of which can change, send or delete something, and 3 findings (1 high) from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.

Can the PubSpace MCP server delete data?

Yes. 2 tools can delete, cancel or revoke something: update_page and delete_page.

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

Yes. 1 tool can send, pay, publish or merge: publish_page.

What tools does the PubSpace MCP server have?

12 tools: 2 destructive, 1 irreversible, 6 write, 3 read. update_page, delete_page, publish_page, unpublish_page, resolve_comment, set_comments, create_account, verify_email and 4 more.

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

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

Does the PubSpace 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): publish_page and set_comments.

Has the PubSpace MCP server changed its tools?

Yes, once since Anlyon Radar started tracking it, most recently on Nov 15, 2026. Every change is on this page and in its Atom feed.