"n8n in Practice 2026 — Wiring Up Discord, Notion, and Gmail"

From OAuth setup to your first three working workflows, step by step

ํ•ต์‹ฌ ์š”์•ฝ

  • Audience: You read the n8n introduction and want to actually build it. Especially anyone stuck on OAuth.
  • What you'll get: 1) Discord, Notion, and Gmail OAuth/credential setup step by step, 2) three working flows — Discord bot command + AI reply / Gmail → Notion auto-triage / Notion publish → Discord announcement, 3) seven common gotchas + fixes, 4) operational tips (monitoring, errors, backups).
  • Prerequisite: n8n cloud trial or self-hosted install completed; accounts on each service.

1. Pre-flight — three credentials

Where most n8n attempts stall is OAuth/credential setup. Get these once and you're done.

1.1 Discord — Webhook + Bot Token

Discord supports two paths. Webhook for one-way send. Bot Token for two-way (trigger + reply).

Webhook (simple) 1. Discord channel settings → Integrations → Webhooks → New Webhook 2. Copy the URL 3. n8n: Discord node → Credential: paste Webhook URL → Save

Bot Token (two-way) 1. Discord Developer Portal → New Application → Bot → copy Token 2. OAuth2 → URL Generator → scopes: bot, applications.commands → permissions: read/send messages → invite bot to your server with the generated URL 3. n8n: Discord Bot API → paste Bot Token → Save

Most common mistake: pasting the token without inviting the bot to a server → no message events arrive.

1.2 Notion — Internal Integration Token

For Notion, an Internal Integration is far simpler than OAuth.

  1. Notion Integrations → New Integration
  2. Name + workspace → create
  3. Copy Internal Integration Secret
  4. In each page or database you want to automate: ⋯ → Connections → add the integration ← people forget this
  5. n8n: Notion API → paste secret as API Key

Most common mistake: skipping step 4 → "Object not found." The integration must be explicitly attached to each page.

1.3 Gmail — OAuth2

The most finicky. Two options:

Option A: n8n Cloud Managed OAuth2 (easiest) - Cloud users only. Click "Sign in with Google" when adding the credential. Done.

Option B: Self-hosted — set up Google Cloud Console manually 1. Google Cloud Console → create a project 2. APIs & Services → Library → enable Gmail API 3. OAuth consent screen → External → register your email 4. Credentials → Create OAuth 2.0 Client ID → Web application 5. Add https://your-n8n-domain/rest/oauth2-credential/callback to Redirect URIs 6. Copy Client ID / Secret 7. n8n: Gmail OAuth2 → paste Client ID / Secret → click "Connect my account"

Most common mistake: HTTP (not HTTPS) self-host fails OAuth. HTTPS is mandatory — Cloudflare Tunnel or ngrok.

2. Flow 1 — Discord bot command with AI reply

Scenario

Users type /ask <question> in Discord; the bot responds via Claude or GPT.

Steps

1. Discord Trigger node
   - Bot Token credential
   - Trigger on: Message Created
   - Channel: a specific channel ID

2. IF node
   - Condition: message text starts with "/ask "

3. Code or Set node
   - Strip "/ask " prefix → question variable

4. Anthropic Chat node
   - Model: Claude Sonnet 4.6
   - System: "You are a helpful Discord bot. Reply concisely in plain English."
   - User: {{ $json.question }}

5. Discord node (send reply)
   - Operation: Send Message
   - Channel: same channel
   - Reply To: original message ID
   - Content: {{ $json.choices[0].message.content }}

Cost estimate

  • 50 calls/day → 1,500 runs/month → fits Cloud Starter (€24, 2,500 cap).
  • Sonnet 4.6 at ~500 in / 500 out × 1,500 = $2.25 in + $11.25 out ≈ $13.50/month.
  • Total ≈ €24 + $13.50 ≈ €36/month.

To cut cost: route through Haiku for simple Q&A first; escalate to Sonnet only on complex queries.

3. Flow 2 — Gmail → Notion auto-triage

Scenario

Tagged emails are auto-saved to a Notion database with title, sender, summary, link.

Steps

1. Gmail Trigger
   - Credential: Gmail OAuth2
   - Watch label: "Important" or inbox
   - Polling: every 5 minutes (or webhook)

2. Claude Haiku Chat node
   - System: "Summarize the email in one line, max 100 chars."
   - User: {{ $json.snippet }}

3. Notion node
   - Operation: Create Database Page
   - Database: "Email Triage"
   - Properties:
     - Title: {{ $json.subject }}
     - From: {{ $json.from }}
     - Summary: {{ $('Claude Haiku').item.json.choices[0].message.content }}
     - URL: {{ $json.threadUrl }}
     - Date: {{ $json.date }}
     - Source: "Auto"

Optional — add classification

Extend step 2 to classify + summarize:

System: "Classify the email into [Work/Personal/Subscription/Spam]
and add a one-line summary. Output JSON: {category, summary}."

Then add a category Notion property and a Switch node to route different categories to different databases.

Cost

  • 100 emails/day × 30 = 3,000 runs → Cloud Pro €60/month
  • Haiku tokens ≈ $1.50/month
  • Total ≈ €61.50/month

Self-hosted + Haiku → ~$1.50/month total.

4. Flow 3 — Notion publish → Discord announcement

Scenario

A Notion database row flipping to "Status: Published" auto-posts an announcement to a Discord channel.

Steps

1. Schedule Trigger
   - Every 5 minutes

2. Notion node
   - Operation: Database Get Many
   - Database: "Blog Posts"
   - Filter: Status = "Published" AND Last Edited > {{ now - 5min }}

3. Loop Over Items (one row at a time)

4. (Optional) AI Agent node
   - Model: Claude Sonnet 4.6
   - System: "Take the post info and write a Discord announcement under 200 chars.
             Include one emoji. End with the link."
   - User: title + summary + URL

5. Discord node
   - Operation: Send Message
   - Channel: #announcements
   - Content: {{ $('AI Agent').item.json.message }}
   - Embed: title=post title, url=link, image=thumbnail

6. Notion node (again)
   - Operation: Update Page
   - Property: "Discord Announced" = true
   (prevents duplicate posts)

Key trick

  • Step 6 is the de-duplication. Without it, the same page posts every 5 minutes.
  • Or add AND Discord Announced = false to the step 2 filter.

5. Seven common gotchas + fixes

Symptom Cause Fix
Notion "Object not found" Integration not attached to that page Page ⋯ → Connections → add integration
Discord bot doesn't receive messages Bot wasn't invited to the server Use OAuth2 URL Generator to invite
Gmail OAuth "Access blocked" Your email isn't in OAuth consent screen Test Users Add yourself to Test Users
Self-host OAuth redirect fails Not HTTPS or domain mismatch Expose via Cloudflare Tunnel/ngrok; match redirect URI exactly
Same item processed repeatedly No de-duplication flag Add a "Processed" column + filter AND Processed = false
Webhook events never arrive Trial relies on polling Trial → polling; production → webhook
LLM response fails to parse Free-form output "Output JSON only" in system prompt + JSON Parse node

6. Operations tips

6.1 Monitoring

  • Executions tab: failed runs in red — click to see which node failed.
  • Email notifications: Settings → Personal → enable error emails.

6.2 Error workflow

Build a dedicated alert workflow and connect every workflow's "Workflow Settings → Error Workflow."

6.3 Backups

  • Cloud: handled.
  • Self-host: pg_dump (PostgreSQL) or daily SQLite copy. Also export workflow JSON to Git.

6.4 Secret management

  • API keys go in Credentials only — never in node parameters.
  • Self-host: set N8N_ENCRYPTION_KEY env var.

Developer notes

  1. Workflow as code: commit n8n JSON to Git → PR review → import on deploy. Enables CI/CD.
  2. Custom nodes: build your own (docs) for unusual SaaS or internal systems — TypeScript.
  3. Verify webhook secrets: external webhook nodes should validate a query token or HMAC signature. An exposed webhook is open to anyone.
  4. Rate-limit handling: Wait + Loop for token-bucket patterns; or push to an external queue (Redis) + Schedule.
  5. Postgres + multi-instance: run PostgreSQL backend with Worker mode for production throughput.
  6. Tracing: in big workflows, add Set nodes that emit metadata (timestamp, step name) — easy to follow execution logs.

References

This is part 7-2 of 11 in the AI Basics series. Next: Local LLMs — Ollama and LM Studio.

๋Œ“๊ธ€

๋Œ“๊ธ€ ์“ฐ๊ธฐ

์ด ๋ธ”๋กœ๊ทธ์˜ ์ธ๊ธฐ ๊ฒŒ์‹œ๋ฌผ

Agent Memory Engine (2/10) — Building an AI Agent Memory System with SQLite Alone

← 1/10 SQLite Memory Engine f… ๐Ÿ“š Series Index 3/10 memcore Completed → A memory engine that runs without daemons, without dependencies, anywhere ํ•ต์‹ฌ ์š”์•ฝ This post covers how to design an AI agent memory engine (memcore) on a single-file SQLite backend. - Structure: 25 modules, ~6,464 lines, 22 DB tables - Principles: no daemon, single-file portability, no required external dependencies, host-agnostic - Core techniques: 4-layer hybrid search (topic · vector · FTS5 · LIKE), U-tag dialectic-based confidence evolution, ingestion-layer bias rejection rules What This Post Covers One approach to building AI agent memory without a server or cloud vector DB. Specifically: (1) the limitations of text-file-based memory, (2) vector DB alternatives compared and selection criteria, (3) how to layer search tiers, (4) how to evolve confidence scores over time, and (5) how to block biased memory accumulation at the ingestion stage. Design Principles: Why Single-File SQLite Text-file-based ...
Read more »

"ML Foundations (9/9) — PyTorch vs TensorFlow, and the Road to Local LLMs"

์ด๋ฏธ์ง€
← 8/9 Deep Learning Architec… ๐Ÿ“š Series Index (series end) The final part. So far we've focused on models. Now we focus on the tools that actually run them . We'll lay out the philosophical difference between PyTorch and TensorFlow, and trace how Hugging Face transformers , llama.cpp , MLX , and Ollama built the bridge to running large language models on your own machine. By the end you should have the full mental model of "download a pretrained LLM and serve it locally." 0. Learning Objectives Explain the eager-vs-graph difference between PyTorch and TensorFlow. Explain, in graph terms, how autograd automates the backward pass. Use Hugging Face transformers ' AutoModel , AutoTokenizer , and pipeline abstractions. Describe llama.cpp's GGUF quantization, INT4 inference, and CPU-first flow. Describe Apple MLX's use of unified memory and how it differs from PyTorch. Run a local LLM with Ollama and call it through an OpenAI-compatible API. ...
Read more »

"RAG Core Study (14/26) — Evaluation Sets with RAGAS & DeepEval"

์ด๋ฏธ์ง€
Series overview: Series index ๐Ÿ“š Series Index Without an evaluation set, every RAG improvement is a story, not evidence. RAG systems fail in more than one place: retrieval, context selection, answer generation, and grounding. That means "looks better" is not a metric. Part 14 explains how to build a golden dataset , how RAGAS and DeepEval fit into the workflow, and why teams must separate retrieval quality from answer quality if they want tuning decisions to hold up. 0. Prerequisites Part 13 reranking — system quality now depends on multi-stage ranking. Part 1 grounding — a correct-looking answer is not enough. Part 12 Hybrid — multiple retrieval settings need comparison on fixed data. 1. Learning Objectives Build a small but useful golden dataset for RAG. Distinguish retrieval metrics from answer metrics . Use RAGAS and DeepEval in the right roles. Avoid the common traps in judge-model-based evaluation. 2. ํ•ต์‹ฌ ์š”์•ฝ An evaluation set for RAG is ...
Read more »

"ML Foundations (8/9) — Deep Learning Architectures: CNN, RNN, Attention"

์ด๋ฏธ์ง€
← 7/9 Deep Learning Training ๐Ÿ“š Series Index 9/9 PyTorch vs TensorFlow → What did we keep adding on top of an MLP? Vision went CNN, sequences went RNN/LSTM, and eventually both converged on Attention and the Transformer. Each architecture is easier to remember if you read it as what it gave up to gain something else . This part is the one-line summary of the decisive inflection points: LeNet → AlexNet → ResNet → LSTM → Attention → Transformer. 0. Learning Objectives Explain why CNN's convolution, pooling, and weight sharing match images so well. Trace what got unblocked from LeNet → AlexNet → VGG → ResNet as depth grew. State BPTT for RNNs and the vanishing/exploding gradient problem. Write the LSTM gate equations (forget, input, output) with intuition. Write the attention formula in query/key/value form and the scaled dot-product variant. State the precise reasons Transformers replaced RNNs (parallelism + long-range dependencies). 1. ํ•ต์‹ฌ ์š”์•ฝ CNN : weight sharin...
Read more »

"ML Foundations (7/9) — Deep Learning Training: Optimizers, Regularization, Initialization"

← 6/9 Neural Networks ๐Ÿ“š Series Index 8/9 Deep Learning Architec… → Building an MLP, as we did in Part 6, is not the same as getting it to train well . The right optimizer, regularization, initialization, and learning rate — when these line up, deep networks converge. When they don't, the network refuses to learn at all. This part is the catalog of those crafts, with formulas, paper citations, and code in one place. 0. Learning Objectives Compare and write the update rules for SGD, Momentum, Nesterov, and Adam. Explain how Dropout, BatchNorm, and LayerNorm work and where they belong in a model. Derive the variance formulas for Xavier and He initialization and match them to activations. Implement step, cosine, and warmup learning-rate schedules in PyTorch. Explain why gradient clipping is effectively required for RNNs and Transformers. Diagnose the most common training failures (NaN, plateau, overfitting) and apply first-line fixes. 1. ํ•ต์‹ฌ ์š”์•ฝ SGD : \(w \leftarro...
Read more »

OpenClaw to Hermes Migration (2/13) — What to Preserve, Partially Port, or Discard

← 1/13 Current Structure Inve… ๐Ÿ“š Series Index 3/13 What Moves to Hermes → A code review record classifying operational scripts into three tiers: preserve, partial-preserve, and discard. What This Post Covers An asset classification framework (preserve / partial-preserve / discard) for migrating a legacy agent system to a successor A concrete case applying the Strangler Fig + Subset Migration strategy to a memory pipeline Classification results evaluated across LOC, external dependencies, and replaceability axes A method for slimming down a bloated single file (1,310 LOC) using the migration as a natural entry point Problem Definition: Port Everything or Cut Deep? The OpenClaw inventory comprises multiple agents, multiple scheduler/daemon processes, and multiple operational scripts. When moving to the successor system, Hermes, the choice is not binary. A full port carries accumulated technical debt along with it; a minimal port discards hard-won domain logic. Code-revie...
Read more »

AI Agents I Built (5/7) — Building an Automated Blogger API Publishing System

← 4/7 My Life Organizer Agent ๐Ÿ“š Series Index 6/7 Dual → Designing a Markdown-Based Auto-Publish Pipeline with Blogger API v3 + OAuth 2.0 Key Summary Blogger API v3 + OAuth 2.0 enables a pipeline that converts Markdown files into publishable HTML posts blogger_publish.py handles the full flow — frontmatter parsing → body cleanup → HTML conversion → CSS injection → API post — in a single script For bulk publishing, quota management (delay, retry, TOC suppression) is the critical design concern Platform Selection Criteria API access is mandatory for AI agent-driven content publishing. A comparison of platforms: Platform API Cost AdSense Decision WordPress.com REST API Paid plan required Paid only Cost overhead Ghost REST API Self-hosted or paid Manual setup Ops overhead Blogger v3 API Free Native integration Adopted Blogger is free, exposes a REST API, and integrates natively with AdSense. For personal blog automation, this combination i...
Read more »