Understanding the Minimal, Standard, and Full Detail Levels for get_node in n8n-mcp

The get_node tool in n8n-mcp offers three detail levels—minimal (~200 tokens), standard (~1-2K tokens, default), and full (~3-8K tokens)—that control how much node metadata is returned when mode is set to info.

The get_node MCP tool in the czlonkowski/n8n-mcp repository retrieves detailed information about specific n8n nodes. When operating in info mode, the tool uses a detail parameter to optimize token usage and response size across three distinct granularity levels.

Detail Level Breakdown

The three detail levels are defined in src/mcp/tools.ts (lines 77-90) and implemented in src/mcp/server.ts (lines 28-74). Each level corresponds to a specific TypeScript return type and targets different use cases.

Minimal Detail Level

The minimal level returns approximately 200 tokens and includes only basic metadata: nodeType, displayName, description, category, package, plus trigger and webhook flags. This level excludes property lists and version history entirely, making it ideal for quick sanity checks and UI-friendly node lists where low latency is critical.

Standard Detail Level

Standard is the default detail level, consuming roughly 1-2K tokens. It returns essential node schema data including required, common, and optional properties, available operations, and a brief version summary. According to the source code in src/mcp/server.ts, this represents the "AI-friendly" view that most agents need to generate or validate workflow configurations.

Full Detail Level

The full detail level provides a complete node definition at approximately 3-8K tokens. This includes all properties (even rarely-used ones), exhaustive operation lists, comprehensive type-structure metadata, and full version summaries. Use this level for deep debugging, documentation generation, or when the standard view omits required fields.

Source Code Implementation

The detail level logic is implemented across three key files:

• src/mcp/tools.ts (lines 77-90): Declares the get_node tool schema and defines the three detail enum values
• src/mcp/server.ts (lines 27-30 and 28-74): Contains the handleInfoMode switch that maps each detail level to its respective response shape (NodeMinimalInfo, NodeStandardInfo, NodeFullInfo)
• src/mcp/tool-docs/configuration/get-node.ts (lines 20-24): Provides human-readable documentation explaining token costs and best practices

Important: The detail parameter only applies when mode='info'. For version-related modes (versions, compare, breaking, migrations), the detail level is ignored entirely.

Code Examples

Here are practical implementations for each detail level:

// Minimal - Core metadata only (~200 tokens)
await mcpClient.call('get_node', {
  nodeType: 'nodes-base.slack',
  detail: 'minimal',
  mode: 'info',
});
// Returns NodeMinimalInfo

// Standard - Default AI-friendly schema (~1-2K tokens)
await mcpClient.call('get_node', {
  nodeType: 'nodes-base.httpRequest',
  // detail defaults to "standard"
  mode: 'info',
  includeExamples: true,
});
// Returns NodeStandardInfo

// Full - Exhaustive node definition (~3-8K tokens)
await mcpClient.call('get_node', {
  nodeType: 'nodes-base.googleSheets',
  detail: 'full',
  mode: 'info',
});
// Returns NodeFullInfo

Summary

• Minimal (~200 tokens): Basic metadata only (nodeType, displayName, description, category, package) for quick lookups and UI lists.
• Standard (~1-2K tokens, default): Essential schema with properties, operations, and version summaries—optimal for AI workflow generation.
• Full (~3-8K tokens): Complete node definition including all properties and exhaustive metadata for debugging and documentation.
• The detail parameter only functions when mode='info'; other modes ignore this setting.
• Implementation spans src/mcp/tools.ts, src/mcp/server.ts, and src/mcp/tool-docs/configuration/get-node.ts.

Frequently Asked Questions

What happens if I omit the detail parameter in get_node?

When you omit the detail parameter, the tool defaults to standard detail level. As implemented in src/mcp/server.ts, this returns approximately 1-2K tokens containing essential node schema data suitable for most AI-assisted workflow building tasks.

Why does the full detail level return 3-8K tokens while minimal returns only 200?

The full detail level serializes the complete node definition including rarely-used properties, exhaustive type structures, and comprehensive operation metadata from the n8n node source. The minimal level strategically filters this to only core identification fields (nodeType, displayName, etc.) to minimize latency and context window usage.

Can I use detail levels with modes other than info?

No. According to the source code in src/mcp/server.ts, the detail parameter is only respected when mode='info'. For version-related modes like versions, compare, breaking, or migrations, the detail level is ignored because these modes return specialized comparison data rather than general node information.

Which detail level should I use for AI agent workflow generation?

Use the standard detail level (or omit the parameter to accept the default). This level, defined in src/mcp/tools.ts and implemented in src/mcp/server.ts, provides the optimal balance of schema information (~1-2K tokens) including required, common, and optional properties that AI agents need to generate valid n8n workflows without excessive context window consumption.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}