> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pnbr.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Text names a tool that your connection does not have

> A connection lists only the tools its access allows, so a message or an entry can name a tool that you cannot call.

Troubleshooting · core

> **Use when**
>
> A Penumbra answer, message or Grammar entry tells you to call an MCP tool that is not in your tool list.

> **Looking for something else?**
>
> A call that is refused for a missing permission (card [Credential](https://docs.pnbr.io/concepts/credential.md)). A call that returned `pending` (problem [A change returned pending](https://docs.pnbr.io/troubleshooting/change-waits-for-approval.md)). A listed tool that fails or contradicts another tool (card [Feedback](https://docs.pnbr.io/concepts/feedback.md)).

## Signs

* An answer says "call `<tool>`", and your connection's tool list has no tool
  with that name.
* A Grammar entry names an MCP tool in its Doors table that you do not have.
* An entry says that the default catalog has no call for the job.

## Why it happens

A connection gets one catalog. The default catalog is small. The full catalog
lists the native tools. A Kit lists the tools that its Kit level and its
Shapes select. A tool also appears only when the connection holds the
permission that it needs, and that access can change. Penumbra writes its
answers for every catalog, so an answer can name a tool that only another
catalog lists.

## What Penumbra does

By itself:

* Gives a connection one catalog: the default catalog, the full catalog or the tools of one Kit.
* Lists a tool only when the connection holds the permission that the tool needs, and follows the current access when it lists them.
* Names a tool in some answers although the connection's catalog does not list it. For example, the default read text can tell an editor to answer a waiting citation with `evidence_review`.

You must:

* Read your tool list before you plan a job around a tool.
* Use another door for the job, or ask for a different connection.
* Tell the person when a job needs a connection or a permission that you do not have.

Not handled:

* The answer does not say which catalog lists the tool.

## Do

1. Refresh the tool list of your connection. Tool availability follows the
   current access, and a new client session starts with the current list.
2. Find the job in its card. The Doors table names the CLI command, the MCP
   tool for each catalog, and the SDK call. Use the door that you have.
3. If the card says that the default catalog has no call for the job, use the
   CLI or the SDK. If you have neither, tell the person that the job needs the
   full catalog or a Kit that selects the tool.
4. If a tool needs a permission that you do not hold, tell the person which
   role or scope is missing (problem [what-can-this-credential-see](https://docs.pnbr.io/troubleshooting/what-can-this-credential-see.md)). Do not
   call a tool that is not listed.
5. If the text names a tool that no catalog lists, send a feedback report. Do
   the same if the Doors table of the card does not match the tool list (card
   rules [feedback.1](https://docs.pnbr.io/concepts/feedback.md#rule-feedback-1) and [feedback.2](https://docs.pnbr.io/concepts/feedback.md#rule-feedback-2)). Name the text and the tool.

## Don't

* Don't call a tool that your list does not show, or invent a call that does the
  same job.
* Don't tell the person that Penumbra cannot do the job. Another door may do it.
* Don't send feedback for a permission that is missing. That is a refusal that
  the person can fix (card rule [feedback.1](https://docs.pnbr.io/concepts/feedback.md#rule-feedback-1)).

## Check

* You named the door that does the job: a listed tool, a CLI command or an SDK
  call.
* The person knows if the job needs another connection.

## Related

Card [credential](https://docs.pnbr.io/concepts/credential.md). Card [project](https://docs.pnbr.io/concepts/project.md). Card [feedback](https://docs.pnbr.io/concepts/feedback.md), with rules [feedback.1](https://docs.pnbr.io/concepts/feedback.md#rule-feedback-1) and
[feedback.2](https://docs.pnbr.io/concepts/feedback.md#rule-feedback-2). Card [approval](https://docs.pnbr.io/concepts/approval.md). Card [workspace](https://docs.pnbr.io/concepts/workspace.md). Card [extraction](https://docs.pnbr.io/concepts/extraction.md). Problem
[what-can-this-credential-see](https://docs.pnbr.io/troubleshooting/what-can-this-credential-see.md).

## Keep going

[Credential](https://docs.pnbr.io/concepts/credential.md) · [Project](https://docs.pnbr.io/concepts/project.md) · [Feedback](https://docs.pnbr.io/concepts/feedback.md) · [Approval](https://docs.pnbr.io/concepts/approval.md) · [Workspace](https://docs.pnbr.io/concepts/workspace.md) · [Extraction](https://docs.pnbr.io/concepts/extraction.md)

Penumbra Grammar 0.6.26


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.