HoldKit

ai.peerfoundry/holdkitpeerfoundry.aistreamable-httpv0.1.0Scanned 1 hour ago

Idempotency keys, locks, timers, and inboxes for agents over MCP. No payments.

14 tools: 4 can delete or cancel things, 7 change things and 3 only read.

Reads private data · 2
Reads text others write · 2
Can send data out · 1
No findings

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 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 →

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

Deletes or cancels 4Removes, cancels or revokes something.Claude Code: deny

inbox.ackMark inbox events done.No findings
Why it is in this group
- The description's first verb is "mark": a word for a change.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it
Mark inbox events done.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
kv.deleteDelete a value.No findings
Why it is in this group
- The name "kv.delete" says "delete": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it
Delete a value.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
lock.releaseRelease a lock with its fencing token.No findings
Why it is in this group
- The name "lock.release" says "lock": a word for a change.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it
Release a lock with its fencing token.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
timer.cancelCancel a timer.No findings
Why it is in this group
- The name "timer.cancel" says "cancel": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it
Cancel a timer.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true

Changes things 7Creates or changes something that can be changed back.Claude Code: ask

idem.beginStart an idempotency key or return the prior outcome.No findings
Why it is in this group
- The description's first verb is "start": a word for a change.
Description, as the server wrote it
Start an idempotency key or return the prior outcome. fingerprint is a hex hash.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
idem.commitStore the outcome reference for an idempotency key.No findings
Why it is in this group
- The name "idem.commit" says "commit": a word for a change.
Description, as the server wrote it
Store the outcome reference for an idempotency key. scope (max 64) and owner are optional. When owner is omitted, the caller key hash is sent.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
inbox.createCreate an inbox.No findings
Why it is in this group
- The name "inbox.create" says "create": a word for a change.
Description, as the server wrote it
Create an inbox. Returns webhook and A2A URLs.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
keys.createMint a pf_ caller key bound to one HoldKit namespace.No findings
Why it is in this group
- The name "keys.create" says "create": a word for a change.
Description, as the server wrote it
Mint a pf_ caller key bound to one HoldKit namespace.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
kv.setWrite a small value with a TTL.No findings
Why it is in this group
- The name "kv.set" says "set": a word for a change.
Description, as the server wrote it
Write a small value with a TTL. Emails and card numbers are rejected.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
lock.acquireTake a lease lock.No findings
Why it is in this group
- The name "lock.acquire" says "lock": a word for a change.
Description, as the server wrote it
Take a lease lock. Returns a fencing token.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
timer.setSchedule a durable timer.Can send data outNo findings
Why it is in this group
- The name "timer.set" says "set": a word for a change.
What it can reach
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it
Schedule a durable timer. Delivery URLs must be public https.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false

Only reads 3Looks things up and changes nothing.Show 3 read-only tools

inbox.pollRead inbox events after a cursor.Reads private dataReads text others writeNo findings
Why it is in this group
- The description's first verb is "read": a word for reading.
What it can reach
- private-dataThe name "inbox.poll" says "inbox": it reads mail.
- private-dataThe description's first sentence says "read" and "inbox": it reads mail.
- private-dataThe input names an inbox (input.inbox_id), and the tool reads: it reads mail.
- untrusted-inputThe name "inbox.poll" says "inbox": it reads inbound email, which other people can write.
- untrusted-inputThe description's first sentence says "read" and "inbox": it reads inbound email, which other people can write.
Description, as the server wrote it
Read inbox events after a cursor.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
inbox.waitPoll an inbox.Reads private dataReads text others writeNo findings
Why it is in this group
- The name "inbox.wait" says "wait": a word for reading.
What it can reach
- private-dataThe name "inbox.wait" says "wait" and "inbox": it reads mail.
- private-dataThe description's first sentence says "inbox": it reads mail.
- private-dataThe input names an inbox (input.inbox_id), and the tool reads: it reads mail.
- untrusted-inputThe name "inbox.wait" says "wait" and "inbox": it reads inbound email, which other people can write.
- untrusted-inputThe description's first sentence says "inbox": it reads inbound email, which other people can write.
Description, as the server wrote it
Poll an inbox. block_ms is accepted and capped at 0 in this MVP, so the call returns current events.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
kv.getRead a small value.No findings
Why it is in this group
- The name "kv.get" says "get": a word for reading.
Description, as the server wrote it
Read a small value.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: 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.

4 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 data2: Reads files, mail, messages, databases, secrets, calendars or repositories.; inbox.poll
inbox.wait
Reads text others write2: Brings in text someone else can write: web pages, inbound email, messages, issues and comments.; inbox.poll
inbox.wait
Can send data out1: Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.; timer.set

deny to break every path

{
  "permissions": {
    "deny": [
      "mcp__holdkit__timer_set"
    ]
  }
}

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

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.

No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.

All changes to this server

Install

Pinned to version 0.1.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http holdkit https://holdkit.peerfoundry.ai/mcp

claude mcp add --transport http holdkit https://holdkit.peerfoundry.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__holdkit__inbox_poll",
      "mcp__holdkit__inbox_wait",
      "mcp__holdkit__kv_get"
    ],
    "ask": [
      "mcp__holdkit__keys_create",
      "mcp__holdkit__inbox_create",
      "mcp__holdkit__timer_set",
      "mcp__holdkit__lock_acquire",
      "mcp__holdkit__idem_begin",
      "mcp__holdkit__idem_commit",
      "mcp__holdkit__kv_set"
    ],
    "deny": [
      "mcp__holdkit__inbox_ack",
      "mcp__holdkit__timer_cancel",
      "mcp__holdkit__lock_release",
      "mcp__holdkit__kv_delete"
    ]
  }
}

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 holdkit. Review it before you use it.

Capability tags behind 3 rules

mcp__holdkit__inbox_poll private-data, untrusted-input
mcp__holdkit__inbox_wait private-data, untrusted-input
mcp__holdkit__timer_set outbound

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://anlyon.dev/api/v1/servers/ai.peerfoundry/holdkit/verify?hash=75fca02b7a8522d38a4abad6593c8b1464010168b87bf7fd05a8baf05b7a72cd" | jq -e '.changed == false'

Pinned hash 75fca02b7a85, from Oct 31, 2026, 11:17 UTC.

Watch

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

https://anlyon.dev/servers/ai.peerfoundry/holdkit/changes.xml

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

Badge

[![anlyon.dev](https://anlyon.dev/badge/ai.peerfoundry/holdkit.svg)](https://anlyon.dev/servers/ai.peerfoundry/holdkit)

Source: MCP Registry. Synced Oct 31, 2026, 11:17 UTC.