lab/nusquama-n8nworkflows-xyz
1
0
Fork 0
mirror of https://github.com/nusquama/n8nworkflows.xyz.git synced 2025-11-27 01:35:37 +08:00
Code Issues Packages Projects Releases Wiki Activity

creation

Browse Source
This commit is contained in:
nusquama
2025-11-20 09:25:23 +01:00
parent 97b65a0d9e
commit 9b4a19628b
1 changed files with 374 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

374
workflows/Daily Google Tasks Briefing in Slack with Ollama-Powered Summaries-8354/readme-8354.md Normal file
View File

@@ -0,0 +1,374 @@
Daily Google Tasks Briefing in Slack with Ollama-Powered Summaries
https://n8nworkflows.xyz/workflows/daily-google-tasks-briefing-in-slack-with-ollama-powered-summaries-8354
# Daily Google Tasks Briefing in Slack with Ollama-Powered Summaries
### 1. Workflow Overview
This workflow, titled **"Daily Google Tasks Briefing in Slack with Ollama-Powered Summaries"**, is designed to automate the delivery of a concise, actionable morning briefing of Google Tasks due on the current day, posted directly into a Slack channel. It targets users who want an efficient daily summary of their tasks without manual review, leveraging a local large language model (LLM) via Ollama to generate human-friendly text summaries.
**Use cases:**
- Daily executive or personal task management briefing
- Automated team updates on task progress
- Reducing manual effort in summarizing task data
**Logical blocks:**
- **1.1 Scheduled Trigger**: Initiates the workflow every morning at 7:00 AM.
- **1.2 Google Tasks Retrieval**: Fetches all tasks from a selected Google Tasklist.
- **1.3 Filtering Tasks Due Today**: Filters tasks to keep only those due on the current day in a specified timezone.
- **1.4 Conditional Branching**: Determines if tasks exist for today and routes accordingly.
- **1.5 LLM Prompt Construction**: Builds a compact, Markdown-formatted prompt for the LLM summarization.
- **1.6 LLM Processing**: Sends the prompt to an Ollama-powered LangChain LLM to generate the briefing.
- **1.7 Output Cleanup**: Cleans LLM output by removing any internal `<think>` tags.
- **1.8 Slack Notification**: Posts the final briefing or a "no tasks" message to a configured Slack channel.
---
### 2. Block-by-Block Analysis
---
#### 1.1 Scheduled Trigger
**Overview:**
Starts the workflow daily at 7:00 AM to automate task briefing generation.
**Nodes Involved:**
- Trigger at Morning
- Sticky Note (Scheduled Trigger)
**Node Details:**
- **Trigger at Morning**
- Type: Cron Trigger
- Configuration: Fires daily at hour 7 (7:00 AM) local time.
- Input: None (trigger)
- Output: Starts the workflow chain.
- Failure types: Cron misconfiguration or system time errors may cause missed triggers.
- Version: n8n Cron node v1
- Sub-workflow: None
- **Sticky Note (Scheduled Trigger)**
- Type: Sticky Note
- Purpose: Documentation only, describing the scheduled trigger’s purpose.
- No inputs or outputs.
---
#### 1.2 Google Tasks Retrieval
**Overview:**
Retrieves all tasks from a specific Google Tasklist using Google Tasks API credentials.
**Nodes Involved:**
- Get many tasks
- Sticky Note (Tasks Briefing)
**Node Details:**
- **Get many tasks**
- Type: Google Tasks node
- Operation: `getAll` retrieves all tasks from the specified Tasklist ID (`MTQzNzI1NzAyOTQxNTk3NzM5NDc6MDow`).
- Returns all tasks without pagination limit.
- Credentials: Requires valid Google Tasks OAuth2 API credentials with read or read/write scope.
- Input: Trigger from Cron node
- Output: Array of task objects (JSON)
- Failure types: OAuth token expiration, insufficient API permissions, quota limits, or invalid tasklist ID.
- Version: Google Tasks node v1
- Sub-workflow: None
- **Sticky Note (Tasks Briefing)**
- Contains comprehensive documentation about the workflow purpose and setup instructions.
- No inputs or outputs.
---
#### 1.3 Filtering Tasks Due Today
**Overview:**
Filters the retrieved tasks to include only those due today, using timezone-aware date comparison for Asia/Dhaka timezone.
**Nodes Involved:**
- Code (Filter Due Today)
- If
- Slack1 (No tasks notification)
- Sticky Note1 (Filter Tasks)
**Node Details:**
- **Code (Filter Due Today)**
- Type: Code node (JavaScript)
- Function:
- Converts current date and task due dates into Asia/Dhaka timezone.
- Filters out tasks with statuses `completed`, `hidden`, or `deleted`.
- Emits individual items for tasks due today.
- If none found, emits a single item `{ __empty: "empty" }`.
- Inputs: Array of all tasks from Google Tasks node
- Outputs: Array of tasks due today or a single empty flag object
- Failure types: Date parsing errors, timezone misconfiguration, empty task arrays.
- Version: Code node v2
- **If**
- Type: If node
- Condition: Checks if the input JSON does not equal `{ __empty: "empty" }`.
- Routes:
- True: There are tasks due today → continue to LLM processing
- False: No tasks due today → send notification to Slack1 node
- Inputs: Output from Code (Filter Due Today)
- Outputs: Two branches (true/false)
- Failure types: Expression evaluation errors or unexpected input structure
- Version: If node v2
- **Slack1**
- Type: Slack node
- Purpose: Sends a message saying "🌤️ No Google Tasks due today. I’ll check again tomorrow" when no tasks are due.
- Configuration:
- Channel: Must be replaced with actual Slack channel name.
- Text: Static message.
- Credentials: Slack API with proper scopes (`chat:write`).
- Input: False branch from If node
- Output: None
- Failure types: Slack API errors, invalid channel, token expiration
- Version: Slack node v2
- **Sticky Note1 (Filter Tasks)**
- Documents filtering logic that routes tasks due today to LLM or notifies absence.
- No inputs or outputs.
---
#### 1.4 LLM Prompt Construction
**Overview:**
Transforms the filtered tasks into a concise, Markdown-formatted prompt suitable for the LLM to generate a Slack-ready briefing.
**Nodes Involved:**
- Code (Build LLM Prompt)
- Sticky Note2 (Process Tasks Briefing)
**Node Details:**
- **Code (Build LLM Prompt)**
- Type: Code node (JavaScript)
- Function:
- Maps each task to a simplified object containing title, notes, due/updated dates, links, and parent info.
- Constructs instructions for the LLM to produce a Slack-ready summary with specific formatting rules.
- Embeds the JSON data as context for the LLM prompt.
- Output is a single JSON object with property `input` containing the full prompt string.
- Inputs: Tasks due today from If node (true branch)
- Outputs: Single prompt item for LLM node
- Failure types: JSON serialization issues, missing task fields, malformed input data
- Version: Code node v2
- **Sticky Note2 (Process Tasks Briefing)**
- Describes building the briefing and cleaning output from internal tags.
- No inputs or outputs.
---
#### 1.5 LLM Processing
**Overview:**
Uses LangChain and the Ollama local LLM model to generate a concise text briefing based on the prompt.
**Nodes Involved:**
- Ollama Model1
- Basic LLM Chain1
**Node Details:**
- **Ollama Model1**
- Type: LangChain Ollama LLM node
- Configuration:
- Model: `qwen3:4b` (a 4-billion parameter Qwen3 model)
- Credentials: Ollama API credentials pointing to local Ollama instance
- Input: Receives prompt from Basic LLM Chain1 (ai_languageModel input)
- Output: LLM-generated text response
- Failure types: Ollama not running, model not available, connection refused, timeout
- Version: LangChain Ollama node v1
- **Basic LLM Chain1**
- Type: LangChain LLM Chain node
- Configuration:
- Uses the prompt text from Code (Build LLM Prompt) node as `text` input
- No batching configured
- Prompt type set to "define" to handle the custom prompt format
- Input: Receives prompt JSON from Code (Build LLM Prompt)
- Output: Sends prompt text to Ollama Model1, receives generated text
- Failure types: LangChain errors, prompt misconfiguration, input data errors
- Version: LangChain Chain LLM node v1.7
---
#### 1.6 Output Cleanup
**Overview:**
Sanitizes the LLM output by removing any internal `<think>...</think>` tags, which may be used by the LLM to represent internal reasoning not meant for display.
**Nodes Involved:**
- Code (Cleanup)
- Sticky Note2 (also documents this step)
**Node Details:**
- **Code (Cleanup)**
- Type: Code node (JavaScript)
- Function:
- Extracts the `text` field from LLM output JSON.
- Uses a regex to remove all occurrences of `<think>...</think>` blocks (case-insensitive, multiline).
- Trims the cleaned text and returns it in `cleanText` property.
- Input: Output from Basic LLM Chain1 (LLM text)
- Output: Cleaned text ready for Slack message
- Failure types: Missing or malformed text field, regex errors
- Version: Code node v2
---
#### 1.7 Slack Notification
**Overview:**
Posts the cleaned task briefing message into the configured Slack channel.
**Nodes Involved:**
- Send a message1
- Sticky Note3 (Notify in Slack)
**Node Details:**
- **Send a message1**
- Type: Slack node
- Configuration:
- Text: Uses the cleaned text from Code (Cleanup) node (`{{ $json.cleanText }}`).
- Channel: Must be replaced with the actual Slack channel name (e.g., `#general`).
- Credentials: Slack API with `chat:write` permission.
- Input: Cleaned LLM output from Code (Cleanup) node
- Output: None (sends message externally)
- Failure types: Slack API errors, invalid channel or token, network issues
- Version: Slack node v2.3
- **Sticky Note3 (Notify in Slack)**
- Documents the purpose of sending the briefing message to Slack.
- No inputs or outputs.
---
### 3. Summary Table
| Node Name | Node Type | Functional Role | Input Node(s) | Output Node(s) | Sticky Note |
|------------------------|---------------------------------|-------------------------------------------|-----------------------------|------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------|
| Trigger at Morning | Cron Trigger | Scheduled daily trigger at 7:00 AM | None | Get many tasks | ## Scheduled Trigger: Every Morning at 7 AM Fetch all the tasks |
| Get many tasks | Google Tasks | Retrieve all tasks from specified tasklist| Trigger at Morning | Code (Filter Due Today) | ## Tasks Briefing (shared with other nodes) |
| Code (Filter Due Today) | Code (JavaScript) | Filter tasks due today in Asia/Dhaka timezone| Get many tasks | If | ## Filter Tasks: Send tasks due today to LLM or notify user that there are no tasks scheduled today |
| If | If | Branch: tasks exist or not | Code (Filter Due Today) | Code (Build LLM Prompt) / Slack1| ## Filter Tasks (see above) |
| Slack1 | Slack | Notify no tasks due today | If (false branch) | None | ## Filter Tasks (see above) |
| Code (Build LLM Prompt)| Code (JavaScript) | Build concise Markdown prompt for LLM | If (true branch) | Basic LLM Chain1 | ## Process Tasks Briefing: Produce a solid briefing for the user with LLM and clean the output removing the Think tags |
| Basic LLM Chain1 | LangChain LLM Chain | Send prompt to Ollama LLM and get response| Code (Build LLM Prompt) | Code (Cleanup) | ## Process Tasks Briefing (see above) |
| Ollama Model1 | LangChain Ollama LLM | Local LLM model for text generation | Basic LLM Chain1 (ai_languageModel) | Basic LLM Chain1 (LLM output) | ## Process Tasks Briefing (see above) |
| Code (Cleanup) | Code (JavaScript) | Remove <think> tags from LLM output | Basic LLM Chain1 | Send a message1 | ## Process Tasks Briefing (see above) |
| Send a message1 | Slack | Post final briefing message to Slack | Code (Cleanup) | None | ## Notify in Slack: Send the task briefing as a slack message |
| Sticky Note | Sticky Note | Documentation | None | None | ## Scheduled Trigger (see above) |
| Sticky Note1 | Sticky Note | Documentation | None | None | ## Filter Tasks (see above) |
| Sticky Note2 | Sticky Note | Documentation | None | None | ## Process Tasks Briefing (see above) |
| Sticky Note3 | Sticky Note | Documentation | None | None | ## Notify in Slack (see above) |
| Sticky Note7 | Sticky Note | Full workflow documentation & instructions| None | None | ## Tasks Briefing (full description, setup, credentials, and instructions) |
---
### 4. Reproducing the Workflow from Scratch
1. **Create a Cron node: "Trigger at Morning"**
- Type: Cron Trigger
- Set trigger time: Daily at 7:00 AM (adjust hour as needed).
- No credentials needed.
2. **Create a Google Tasks node: "Get many tasks"**
- Type: Google Tasks
- Operation: `getAll`
- Tasklist ID: Set your Google Tasklist ID (example: `MTQzNzI1NzAyOTQxNTk3NzM5NDc6MDow`).
- Credentials: Set up Google Tasks OAuth2 API credentials with scope `https://www.googleapis.com/auth/tasks.readonly` (or read/write).
- Connect Trigger at Morning → Get many tasks
3. **Create a Code node: "Code (Filter Due Today)"**
- Type: Code (JavaScript)
- Paste the provided script to filter tasks due today in timezone `Asia/Dhaka`.
- Adjust `TZ` constant if needed for your timezone.
- Connect Get many tasks → Code (Filter Due Today)
4. **Create an If node: "If"**
- Type: If
- Condition: Check if `$json.__empty` is NOT equal to `"empty"`.
- Connect Code (Filter Due Today) → If
5. **Create a Slack node: "Slack1" (No tasks notification)**
- Type: Slack
- Text: "🌤️ No Google Tasks due today. I’ll check again tomorrow"
- Channel: Replace with your Slack channel name (e.g., `#general`).
- Credentials: Slack API with `chat:write` scope.
- Connect If (false branch) → Slack1
6. **Create a Code node: "Code (Build LLM Prompt)"**
- Type: Code (JavaScript)
- Paste the provided script to build a compact Markdown prompt for LLM from filtered tasks.
- Connect If (true branch) → Code (Build LLM Prompt)
7. **Create a LangChain LLM Chain node: "Basic LLM Chain1"**
- Type: LangChain Chain LLM
- Prompt input: `{{ $json.input }}` from Code (Build LLM Prompt)
- Prompt type: `define`
- No batching
- Connect Code (Build LLM Prompt) → Basic LLM Chain1
8. **Create a LangChain Ollama node: "Ollama Model1"**
- Type: LangChain Ollama LLM
- Model: `qwen3:4b` (or your preferred Ollama model)
- Credentials: Ollama API credentials pointing to your local Ollama instance (default `http://localhost:11434`)
- Connect Ollama Model1 (ai_languageModel output) → Basic LLM Chain1 (ai_languageModel input)
- Connect Basic LLM Chain1 → Ollama Model1 (loop for model call)
9. **Create a Code node: "Code (Cleanup)"**
- Type: Code (JavaScript)
- Paste script to remove `<think>...</think>` blocks from LLM output text.
- Input: Output from Basic LLM Chain1 (LLM text)
- Connect Basic LLM Chain1 → Code (Cleanup)
10. **Create a Slack node: "Send a message1"**
- Type: Slack
- Text: `{{ $json.cleanText }}` from Code (Cleanup)
- Channel: Replace with your Slack channel name
- Credentials: Slack API with `chat:write` scope
- Connect Code (Cleanup) → Send a message1
11. **Verify all connections and credentials:**
- Google Tasks OAuth2 API credentials configured with correct scopes and valid token.
- Slack API credentials set with proper permissions.
- Ollama local server running with the specified model loaded.
- Replace placeholder Slack channel names with actual channel IDs or names.
12. **Test workflow manually:**
- Run workflow manually to confirm retrieval, filtering, summarization, cleanup, and posting work as expected.
- Adjust timezone or schedule settings if necessary.
13. **Activate workflow:**
- Once tested and confirmed, activate the workflow to run daily at scheduled time.
---
### 5. General Notes & Resources
| Note Content | Context or Link |
|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------|
| This workflow posts a clean, Slack-ready morning summary of Google Tasks due today using a local LLM (Ollama) integrated via LangChain. | Workflow purpose |
| Setup requires enabling Google Tasks API in Google Cloud Console and creating OAuth2 credentials with appropriate scopes. | Google Tasks API setup instructions |
| Slack API credentials require bot or user token with `chat:write` scope to send messages to channels. | Slack API setup instructions |
| Ollama must be installed and running locally (`http://localhost:11434` by default), and the desired model (e.g., `qwen3:4b`) must be pulled and available. | Ollama setup and model pull instructions |
| The timezone constant in the filtering code (`Asia/Dhaka`) should be adjusted to your local timezone for accurate daily task filtering. | Timezone adjustment note |
| The workflow includes safety cleanup removing `<think>...</think>` tags from LLM output to avoid internal reasoning snippets leaking to Slack. | Output cleaning best practice |
| Slack channel names must be replaced in the Slack nodes to target your actual communication channel. | Slack channel configuration |
| For further reading and detailed setup, refer to external blog: [https://n8n.io/blog/daily-google-tasks-briefing-with-ollama](https://n8n.io/blog/daily-google-tasks-briefing-with-ollama) | External blog reference |
---
This completes the full structured reference document for the **"Daily Google Tasks Briefing in Slack with Ollama-Powered Summaries"** n8n workflow.