n8n-nodes-analytics

A powerful n8n community node that replaces complex 10–20 node analytics workflows with a single reusable node. It fetches execution data through the n8n API, automatically detects AI/LLM nodes, calculates token usage and cost analytics, and outputs structured results in multiple formats.

Installation · Credentials · Operations · AI Token Analytics · Cost Analytics · Architecture

Features

• 📊 8 operations covering execution, workflow, node, timeline, token, model and cost analytics
• 🤖 Automatic AI node detection — no configuration needed; works with OpenAI, Anthropic, Gemini, Mistral, Groq, Ollama, OpenRouter, Cohere, DeepSeek and any LangChain LLM node
• 🔢 Provider-agnostic token extraction — normalizes input_tokens/output_tokens, prompt_tokens/completion_tokens, cached and reasoning tokens
• 💰 Cost estimation with a built-in pricing table, fuzzy model matching, custom pricing overrides and multi-currency output
• 📤 Multiple export formats — JSON, flat JSON, CSV, Markdown
• ⚡ Single-pass, Map-based aggregation — one traversal of runData, lazy evaluation per operation

Installation

Community Nodes (recommended)

1. In n8n, go to Settings → Community Nodes.
2. Select Install.
3. Enter n8n-nodes-analytics and confirm.

Manual / npm

npm install n8n-nodes-analytics

For local development:

git clone <this-repo>
cd n8n-nodes-analytics
npm install
npm run build
# link into your n8n custom nodes directory

Credentials

The node authenticates against the n8n REST API using an n8n API credential.

1. In n8n, open Settings → API and create an API key.
2. In the credential, set:
   • Base URL — your n8n instance URL, e.g. http://localhost:5678 (no trailing slash)
   • API Key — the key you just created

The credential is verified automatically against GET /api/v1/executions.

The API key is sent as the X-N8N-API-KEY header. All requests go to {{baseUrl}}/api/v1/....

Operations

Select the Execution Analytics resource, then one of the following operations. Every operation accepts an Execution ID, or enable Use Current Execution to analyze the running execution.

Advanced Options (collection)

Toggle what appears in the output:

• Include Token Usage
• Include Costs
• Include Errors
• Include Node Timing
• Include Workflow Metadata
• Include AI Metadata
• Include Raw Execution (attaches the full raw execution — large)

AI Token Analytics

AI nodes are detected two ways, so the node keeps working as new integrations appear:

1. Node type matching — known LangChain / base LLM node types (lmChatOpenAi, lmChatAnthropic, lmChatGoogleGemini, lmMistral, lmChatGroq, lmOllama, lmChatOpenRouter, …) and any type containing langchain, lm, ai, agent, embeddings, or a provider name.
2. Output inspection (future-proof fallback) — any node whose output contains a tokenUsage / usage / token_usage object (found recursively) is treated as an AI node.

Token fields are normalized across providers:

Provider support matrix

Cost Analytics

Costs are estimated from the built-in pricing table (USD per 1M tokens), with fuzzy prefix matching so versioned model names like gpt-4o-2024-08-06 resolve to gpt-4o.

(abbreviated — see Helpers.ts for the full table)

• Cached tokens are billed at 50% of the prompt rate and excluded from the full-price prompt total.
• Custom pricing (Cost Source → Custom Pricing) is merged over the built-in table:

  { "my-model": { "prompt": 1.0, "completion": 2.0 } }
  

• Currency — costs are computed in USD then converted (USD, EUR, NPR, GBP, INR) for display convenience.

Architecture

flowchart TD
    A[Analytics.node.ts<br/>execute] --> B[GenericFunctions<br/>resolveExecutionId]
    B --> C[Validators<br/>validateExecutionId]
    A --> D[Transport<br/>fetchExecution]
    D --> E[GenericFunctions<br/>n8nApiRequest] --> F[(n8n REST API)]
    A --> G[Validators<br/>validateExecutionData]
    A --> H{Operation router}
    H --> I[Helpers<br/>extractNodeAnalytics]
    H --> J[Helpers<br/>extractAiAnalytics]
    J --> K[detectProvider / extractTokenUsage / extractModel]
    H --> L[Helpers<br/>aggregateTokens]
    H --> M[Helpers<br/>calculateCost + PRICING]
    H --> N[Helpers<br/>buildTimeline / filterNodes]
    H --> O[Helpers<br/>toFlatJson / toCsv / toMarkdown]
    I --> P[INodeExecutionData]
    J --> P
    L --> P
    M --> P
    N --> P
    O --> P

Development

npm install
npm run lint      # ESLint (TypeScript)
npm run build     # tsc + copy icons to dist
npm test          # Jest unit tests
npm pack          # produce the publishable tarball

Unit tests cover token extraction across provider formats, provider detection, cost calculation (built-in + custom + currency + cached discount), aggregation by every group-by mode, export formatters, filters, timeline ordering, and end-to-end extraction over an execution fixture.

Changelog

1.0.0

• Initial release: 8 operations, automatic AI node detection, token normalization, cost estimation with custom pricing & multi-currency, and JSON/flat-JSON/CSV/Markdown export.

License

MIT