Daily Google Tasks Briefing in Slack with Ollama-Powered Summaries

Description

Tasks Briefing

This template posts a clean, Slack-ready morning summary of your Google Tasks due today. It fetches tasks, filters only those due “today” in your timezone, asks a local LLM (via LangChain + Ollama) to produce a short summary (no steps, just a concise brief), strips any hidden <think> blocks, and delivers the message to your chosen Slack channel.

How it works

Trigger at Morning (Cron) – runs at 7:00 AM (you can change the hour) to kick things off daily.
Get many tasks (Google Tasks node) – pulls tasks from your selected Google Tasklist.
Code (Filter Due Today) – normalizes dates to your timezone, keeps only tasks due today, and emits a fallback flag if none exist.
If – routes:

True (has tasks) → continues to the LLM summary path.
False (no tasks) → sends a “No tasks due today” message to Slack.
Code (Build LLM Prompt) – builds a compact, Markdown-only prompt for the model (no tool calls).
Basic LLM Chain (LangChain) + Ollama Model – generates a short summary for Slack.
Code (Cleanup) – removes any <think>…</think> content if the model includes it.
Send a message (Slack) – posts the final brief to your Slack channel.

Required credentials

Google Tasks OAuth2 API** – to read tasks from your Google Tasklist.
Slack API** – to post the summary into a channel.
Ollama** – local model endpoint (e.g., qwen3:4b); used by the LangChain LLM nodes.

Setup Instructions

Google Tasks credential

In Google Cloud Console: enable Google Tasks API, create an OAuth Client (Web), and set the redirect URI shown by n8n.
In n8n Credentials, add Google Tasks OAuth2 API with scope:

https://www.googleapis.com/auth/tasks (read/write) or
https://www.googleapis.com/auth/tasks.readonly (read-only).
In the Get many tasks node, select your credential and your Tasklist.

Slack credential & channel

In n8n Credentials, add Slack API (bot/user token with chat:write).
In Send a message nodes, select your Slack credential and set the Channel (e.g., #new-leads).

Ollama model (LangChain)

Ensure Ollama is running on your host (default http://localhost:11434).
Pull a model (e.g., ollama pull qwen3:4b) or use another supported model (llama3:8b, etc.).
In Ollama Model node, select your Ollama credential and set the model name to match what you pulled.

Timezone & schedule

The Cron node is set to 7:00 AM. Adjust as needed.
The Code (Filter Due Today) node is configured for Asia/Dhaka; change the TZ constant if you prefer a different timezone.

(Optional) Cleanup safety

The template includes a Code (Cleanup) node that strips <think>…</think> blocks from model output. Keep this connected before the Slack node.

Test the flow

Run the workflow once manually:

If you have tasks due today, you should see a concise summary posted to your Slack channel.
If none are due, you’ll receive a friendly “No tasks due today” message.

Activate

When everything looks good, toggle the workflow Active to receive the daily summary automatically.