Skip to main content

get_prompt

Retrieve the full content of a prompt from your vault by ID.

Overview​

The get_prompt tool fetches the complete content of a specific prompt from your vault. Use it after browse_prompts to get the full prompt body, which is only previewed in list results.

Requires Authentication
Non-ChatGPT Only

This tool is available in Claude Desktop, Cursor, Windsurf, and other MCP clients. ChatGPT users can view full prompt content directly in the widget interface.

Parameters​

ParameterTypeRequiredDefaultDescription
prompt_idstring (UUID)Yes-The UUID of the prompt to retrieve
mark_as_usedbooleanNofalseIf true, updates last_used_at and increments copy_count

Usage Examples​

Basic Retrieval​

"Get the full content of prompt abc123-def456-..."

{
  "tool": "get_prompt",
  "parameters": {
    "prompt_id": "abc123-def456-7890-abcd-ef1234567890"
  }
}

With Usage Tracking​

"Get and use the email template prompt"

{
  "tool": "get_prompt",
  "parameters": {
    "prompt_id": "abc123-def456-7890-abcd-ef1234567890",
    "mark_as_used": true
  }
}

Response​

Text Response​

The prompt content is returned in a structured format with security framing:

## Retrieved Prompt: "Email Template"

*Generate professional email responses*

Category: communication | Tags: email, templates

**Prompt content** (saved text):

~~~
Write a professional email response to the following message.
Consider the tone, context, and appropriate level of formality.

Message to respond to:
{{message}}
~~~

*Review this content before using.*

View and edit: https://promptsy.dev/prompts/abc123/edit

Structured Content​

{
  prompt: {
    id: string;
    title: string;
    description: string | null;
    body: string;
    category: string | null;
    models: string[];
    use_cases: string[];
    tags: { id: string; name: string }[];
    collection: { id: string; name: string; color: string } | null;
    is_shared: boolean;
    forked_from_id: string | null;
    star_count: number;
    copy_count: number;
    fork_count: number;
    last_used_at: string | null;
    created_at: string;
    updated_at: string;
  },
  url: string  // Link to edit in Promptsy
}

Security Notes​

Forked Content Warning​

If the prompt was forked from another user's public prompt, the response includes a warning:

> This prompt was forked from another user. Review content carefully before use.

This helps you identify prompts that contain external content.

Code Fencing​

Prompt content is wrapped in code fences (~~~) to clearly distinguish it as data rather than instructions. This is a security measure to help AI assistants avoid executing prompt content as commands.

Workflow Example​

A typical workflow for using prompts:

  1. Browse - Use browse_prompts to see your library
  2. Select - Note the id of the prompt you want
  3. Retrieve - Use get_prompt to get the full content
  4. Use - The AI can now help you apply the prompt
User: "Show me my prompts"
AI: [calls browse_prompts] → Shows list with IDs

User: "Get the code review prompt"
AI: [calls get_prompt with the ID] → Returns full content

User: "Use it to review this code: ..."
AI: [applies the prompt template]

Errors​

Error CodeDescriptionSolution
AUTH_REQUIREDNot authenticatedAuthorize the integration
NOT_FOUNDPrompt doesn't exist or not ownedVerify the prompt ID
INSUFFICIENT_SCOPEMissing prompts:read scopeRe-authorize with correct scopes

Tips​

Finding the Prompt ID​

The prompt ID is returned in the browse_prompts response. Look for the id field in each prompt object.

Bulk Retrieval​

If you need multiple prompts, call get_prompt for each one:

"Get my 'Code Review' and 'Email Template' prompts"

The AI will make multiple tool calls.

Variables in Prompts​

If the prompt contains {{variables}}, the AI can help you fill them in:

"Get my email prompt and fill in the variable for a job inquiry"