Skills

Skills are reusable prompt templates attached to your MCP server. They are exposed to AI clients via the MCP prompts primitive — any MCP-compatible client (Claude, Codex, Cursor, Gemini) sees them automatically alongside your tools.

A client calls prompts/list to enumerate the skills on a server, and prompts/get to retrieve a rendered prompt with its {{argument}} placeholders filled in.

Skill schema

Field	Type	Required	Description
name	string	yes	Unique slug identifier used in prompts/get — e.g. code_review
title	string	no	Human-readable display name shown in the client
description	string	no	Short description of what the skill does
prompt	string	yes	The prompt template. Use {{argument_name}} placeholders for runtime substitution
arguments	array	no	List of argument definitions (see below)

arguments array

Each argument maps to a {{placeholder}} in the prompt template.

Field	Type	Required	Description
name	string	yes	Argument name — must match a {{placeholder}} in the prompt
description	string	yes	Description shown to the agent in prompts/list
required	boolean	no	Whether the argument must be provided. Defaults to false

Example

{
  "name": "summarize_records",
  "title": "Summarize Recent Records",
  "description": "Fetch and summarize records matching a topic from the last 7 days",
  "prompt": "Use the search_records tool to find records matching '{{topic}}' from the last 7 days. Summarize the key themes in 3 bullet points.",
  "arguments": [
    {
      "name": "topic",
      "description": "The topic or keyword to search for",
      "required": true
    }
  ]
}

When an agent calls prompts/get with { "name": "summarize_records", "arguments": { "topic": "billing" } }, it receives:

Use the search_records tool to find records matching 'billing' from the last 7 days. Summarize the key themes in 3 bullet points.

Managing skills

Via the Shredly MCP

You can add and remove skills directly through the Shredly MCP using natural language:

Add a skill named "summarize_records" to my MCP server with prompt:
"Use search_records to find records matching '{{topic}}' from the last 7 days."

How placeholders work

Placeholders follow the {{argument_name}} syntax. At runtime, when an agent calls prompts/get with an arguments map, each placeholder is replaced with the corresponding value.

• Unmatched placeholders (no argument provided) are replaced with an empty string.
• Placeholders are case-sensitive — {{Topic}} and {{topic}} are different.

Skills vs. system prompts

Skills live on the server. When you update a skill, every agent that connects gets the updated version automatically — no client configuration required. System prompts, by contrast, must be configured per client and don't travel with the server.

Use skills to encode the best-practice workflows for your MCP — the prompts you'd otherwise put in a README or a Notion doc.