Plugin

Artifact tools published

Also known as: Extension, Add-on, Module

Definition

A software component that adds functionality to a host application without modifying its core code. In AI systems, plugins typically expose new tools, data sources, or integrations that agents can use. Plugins follow a defined interface contract with the host and can be installed, updated, or removed independently.

What this is NOT

  • Not the same as a skill (plugins provide tools; skills provide knowledge/workflows)
  • Not a standalone application (plugins require a host)
  • Not an API (plugins are installed locally; APIs are called remotely)
  • Not a library (plugins follow a specific interface contract with the host)

Alternative Interpretations

Different communities use this term differently:

chatgpt-plugins

Third-party integrations for ChatGPT that expose APIs the model can call to retrieve information or take actions in external services.

ChatGPT plugins were largely superseded by GPTs and the Assistants API, but established the pattern of LLM-accessible third-party integrations.

Sources: OpenAI ChatGPT Plugins documentation (2023)

ide-extensions

Extensions for code editors like VS Code that add language support, debugging capabilities, AI assistance, or other development features.

Sources: VS Code Extension API documentation, JetBrains Plugin SDK

mcp-ecosystem

MCP servers that expose tools, resources, and prompts to MCP clients, enabling AI applications to interact with external systems through a standardized protocol.

Sources: Model Context Protocol specification

Examples

  • A VS Code extension that provides AI code completion
  • An MCP server that exposes filesystem operations to Claude
  • A browser extension that adds AI summarization to web pages
  • A Slack plugin that enables AI assistants to post messages

Counterexamples

Things that might seem like Plugin but are not:

  • A standalone CLI tool (not attached to a host application)
  • A REST API endpoint (plugins are local; APIs are remote)
  • A skill file (skills provide knowledge, not tool implementations)

Relations

  • overlapsWith skill (Both extend capabilities, but plugins add tools while skills add knowledge)
  • specializes tool-use (Plugins are a mechanism for providing tools to agents)
  • overlapsWith model-context-protocol (MCP is a protocol for plugin-like integrations)

See Also

Implementations

Tools and frameworks that implement this concept: