mirror of
https://github.com/khoaliber/n8nworkflows.xyz.git
synced 2026-04-28 00:29:22 +00:00
creation
This commit is contained in:
nusquama
+293
@@ -0,0 +1,293 @@
|
||||
Automated Blog Generation with Gemini AI, GitHub & Jekyll Publishing
|
||||
|
||||
https://n8nworkflows.xyz/workflows/automated-blog-generation-with-gemini-ai--github---jekyll-publishing-6357
|
||||
|
||||
|
||||
# Automated Blog Generation with Gemini AI, GitHub & Jekyll Publishing
|
||||
|
||||
### 1. Workflow Overview
|
||||
|
||||
This workflow automates the creation and publishing of a technical blog post. It is designed to:
|
||||
|
||||
- Retrieve blog topics from a Google Sheet that tracks blog ideas and their status.
|
||||
- Perform deep research on the selected topic using web search (via Tavily) and Wikipedia content.
|
||||
- Summarize and format the research context.
|
||||
- Use the Google Gemini AI language model (or equivalent LLM) to generate a complete, well-structured blog post in Markdown suitable for Jekyll static site generation.
|
||||
- Prepare and commit the generated Markdown file to a GitHub repository hosting a Jekyll site.
|
||||
- Update the blog topic status in the Google Sheet to “done” upon successful publishing.
|
||||
|
||||
The workflow is organized into the following logical blocks:
|
||||
|
||||
- **1.1 Trigger and Input Retrieval:** Scheduling and fetching blog topics from Google Sheets.
|
||||
- **1.2 Research and Context Gathering:** Using Tavily web search and Wikipedia to collect background information.
|
||||
- **1.3 Content Generation:** Formatting research results and generating the blog post with Gemini AI.
|
||||
- **1.4 File Preparation and Publishing:** Formatting the Markdown file and committing it to GitHub.
|
||||
- **1.5 Status Update:** Marking the blog topic as done in the Google Sheet.
|
||||
|
||||
---
|
||||
|
||||
### 2. Block-by-Block Analysis
|
||||
|
||||
#### 1.1 Trigger and Input Retrieval
|
||||
|
||||
**Overview:**
|
||||
This block initiates the workflow on a schedule, fetches a blog topic from a Google Sheet that is not yet marked as done, and extracts the relevant topic text for processing.
|
||||
|
||||
**Nodes Involved:**
|
||||
- Schedule Trigger
|
||||
- Get Topic from Google Sheet
|
||||
- Extract Blog Title
|
||||
|
||||
**Node Details:**
|
||||
|
||||
- **Schedule Trigger**
|
||||
- Type: Schedule Trigger
|
||||
- Role: Triggers workflow execution daily at 09:00 (configured hour).
|
||||
- Configuration: Set to trigger at 9 AM every day.
|
||||
- Inputs: None (trigger node).
|
||||
- Outputs: Connects to “Get Topic from Google Sheet”.
|
||||
- Edge Cases: Misconfiguration could cause missed triggers; time zone considerations apply (Asia/Kolkata set).
|
||||
|
||||
- **Get Topic from Google Sheet**
|
||||
- Type: Google Sheets
|
||||
- Role: Reads entries from the specified sheet filtered by status not done (status column filter).
|
||||
- Configuration: Reads from sheet named “blog list” (gid=0) in a Google Sheets document. Filters rows where status is not “done”.
|
||||
- Inputs: From Schedule Trigger.
|
||||
- Outputs: Passes data to “Extract Blog Title”.
|
||||
- Credentials: Google Sheets OAuth2.
|
||||
- Edge Cases: Google API quota limits, connectivity issues, empty or malformed rows.
|
||||
|
||||
- **Extract Blog Title**
|
||||
- Type: Set
|
||||
- Role: Extracts the blog topic title field from the sheet data and sets it as a variable `topic`.
|
||||
- Configuration: Assigns `topic` from `$json.Title` (assumed sheet column).
|
||||
- Inputs: From “Get Topic from Google Sheet”.
|
||||
- Outputs: Connects to “Tavily Web Search” for research.
|
||||
- Edge Cases: Missing or empty Title field would cause downstream nodes to fail or generate irrelevant content.
|
||||
|
||||
---
|
||||
|
||||
#### 1.2 Research and Context Gathering
|
||||
|
||||
**Overview:**
|
||||
This block performs deep research on the blog topic by querying Tavily web search and Wikipedia. It aggregates and summarizes the collected information to build context for the AI generation.
|
||||
|
||||
**Nodes Involved:**
|
||||
- Tavily Web Search
|
||||
- Fetch Wikipedia Context
|
||||
- Summarize Search Results
|
||||
|
||||
**Node Details:**
|
||||
|
||||
- **Tavily Web Search**
|
||||
- Type: Tavily Search Node
|
||||
- Role: Performs an advanced web search on the extracted topic to gather relevant articles and summaries.
|
||||
- Configuration: Query set to the extracted `topic` variable, `search_depth` set to “advanced” for richer results.
|
||||
- Inputs: From “Extract Blog Title”.
|
||||
- Outputs: Connects to “Summarize Search Results” and “Google Gemini Chat Model” (via ai_languageModel interface).
|
||||
- Credentials: Tavily API key.
|
||||
- Edge Cases: API rate limits, query failures, zero results.
|
||||
|
||||
- **Fetch Wikipedia Context**
|
||||
- Type: Wikipedia Tool Node (Langchain)
|
||||
- Role: Retrieves Wikipedia content and images related to the topic for enriched context.
|
||||
- Configuration: No parameters explicitly set, uses topic context dynamically.
|
||||
- Inputs: AI Tool input from Tavily results.
|
||||
- Outputs: Feeds into AI agent for blog generation.
|
||||
- Edge Cases: Topic not found on Wikipedia, disambiguation pages, API errors.
|
||||
|
||||
- **Summarize Search Results**
|
||||
- Type: Code Node (JavaScript)
|
||||
- Role: Processes Tavily search results, picks top 3, formats with titles, snippets, and sources into a Markdown-like context string.
|
||||
- Configuration: Custom JS code that creates a concatenated summary with headings and links.
|
||||
- Inputs: From “Tavily Web Search”.
|
||||
- Outputs: Passes summary context and topic to “Generate Blog with Gemini” node.
|
||||
- Edge Cases: Empty search results, malformed data.
|
||||
|
||||
---
|
||||
|
||||
#### 1.3 Content Generation
|
||||
|
||||
**Overview:**
|
||||
This block uses a Gemini AI language model agent to generate a full technical blog post in Markdown format. The prompt includes strict formatting instructions for Jekyll compatibility.
|
||||
|
||||
**Nodes Involved:**
|
||||
- Generate Blog with Gemini
|
||||
|
||||
**Node Details:**
|
||||
|
||||
- **Generate Blog with Gemini**
|
||||
- Type: Langchain Agent Node (Google Gemini AI)
|
||||
- Role: Receives the topic and summarized context; generates a complete Markdown blog post including YAML front matter, table of contents, sections, code snippets, images, and links per detailed prompt instructions.
|
||||
- Configuration:
|
||||
- Prompt text dynamically includes topic and context.
|
||||
- System message sets strict formatting rules for Markdown suitable for Jekyll.
|
||||
- Uses Google Gemini 2.5 Pro model via Google Palm API credentials.
|
||||
- Inputs: From “Summarize Search Results” and “Fetch Wikipedia Context” (via AI tool interface).
|
||||
- Outputs: Produces fully formatted Markdown content.
|
||||
- Credentials: Google Gemini API (Google Palm API).
|
||||
- Edge Cases: API timeouts, rate limits, malformed input causing generation errors, prompt misconfiguration.
|
||||
|
||||
---
|
||||
|
||||
#### 1.4 File Preparation and Publishing
|
||||
|
||||
**Overview:**
|
||||
This block formats the generated Markdown content into a file structure, constructs a commit message, and uploads the file to a specified GitHub repository.
|
||||
|
||||
**Nodes Involved:**
|
||||
- Prepare File for Commit
|
||||
- Commit Blog Post to GitHub
|
||||
|
||||
**Node Details:**
|
||||
|
||||
- **Prepare File for Commit**
|
||||
- Type: Set
|
||||
- Role: Sets file path, content, and commit message for GitHub upload.
|
||||
- Configuration:
|
||||
- Filename format: `YYYY-MM-DD-topic-slug.md` (topic slug is lowercase and spaces replaced by hyphens).
|
||||
- File content: Markdown output from AI agent.
|
||||
- Commit message: “Add blog post: [Topic Title]”.
|
||||
- Inputs: From “Generate Blog with Gemini”.
|
||||
- Outputs: Connects to “Commit Blog Post to GitHub”.
|
||||
- Edge Cases: Incorrect topic formatting causing invalid filenames, empty content.
|
||||
|
||||
- **Commit Blog Post to GitHub**
|
||||
- Type: GitHub Node
|
||||
- Role: Commits the Markdown file to the specified GitHub repository and path (typically `_posts/` folder for Jekyll).
|
||||
- Configuration:
|
||||
- Repository owner and name configured.
|
||||
- File path set dynamically (e.g., `_posts/YYYY-MM-DD-topic-slug.md`).
|
||||
- Commit message as prepared.
|
||||
- Authentication via OAuth2 token.
|
||||
- Inputs: From “Prepare File for Commit”.
|
||||
- Outputs: Connects to “Mark Topic as Done in Sheet”.
|
||||
- Credentials: GitHub OAuth2 credentials.
|
||||
- Edge Cases: GitHub API errors, auth failures, branch conflicts, rate limits.
|
||||
|
||||
---
|
||||
|
||||
#### 1.5 Status Update
|
||||
|
||||
**Overview:**
|
||||
Marks the processed blog topic as done in the Google Sheet to prevent reprocessing.
|
||||
|
||||
**Nodes Involved:**
|
||||
- Mark Topic as Done in Sheet
|
||||
|
||||
**Node Details:**
|
||||
|
||||
- **Mark Topic as Done in Sheet**
|
||||
- Type: Google Sheets
|
||||
- Role: Updates the row corresponding to the blog topic with status “done”.
|
||||
- Configuration:
|
||||
- Matches row by `row_number`.
|
||||
- Sets `status` column to “done”.
|
||||
- Uses same sheet and document as input node.
|
||||
- Inputs: From “Commit Blog Post to GitHub”.
|
||||
- Credentials: Google Sheets OAuth2.
|
||||
- Edge Cases: Row mismatch causing no update, API limits, concurrency conflicts.
|
||||
|
||||
---
|
||||
|
||||
### 3. Summary Table
|
||||
|
||||
| Node Name | Node Type | Functional Role | Input Node(s) | Output Node(s) | Sticky Note |
|
||||
|----------------------------|-----------------------------------------|--------------------------------------------|------------------------------|------------------------------|-----------------------------------------------------------------------------------------------------|
|
||||
| Schedule Trigger | Schedule Trigger | Start workflow on schedule | None | Get Topic from Google Sheet | ## Trigger node<br>Configure the node for your preference or else you can use other option as well. |
|
||||
| Get Topic from Google Sheet | Google Sheets | Retrieve blog topics not marked “done” | Schedule Trigger | Extract Blog Title | ## Get Topic from spreadsheet<br>This will capture the topic which is not done yet. |
|
||||
| Extract Blog Title | Set | Extract topic text from sheet data | Get Topic from Google Sheet | Tavily Web Search | ## Extract topic<br>This will extract topic text from the sheet node. |
|
||||
| Tavily Web Search | Tavily Search | Perform web search for topic research | Extract Blog Title | Summarize Search Results | ## Deep Search about topic<br>We have a system node that gathers information about a given topic. |
|
||||
| Fetch Wikipedia Context | Wikipedia Tool (Langchain) | Retrieve Wikipedia content for topic | AI Tool input from Tavily | Generate Blog with Gemini | |
|
||||
| Summarize Search Results | Code | Summarize and format Tavily search results | Tavily Web Search | Generate Blog with Gemini | ## Format the result<br>Extracts and formats search details into a predefined structure. |
|
||||
| Generate Blog with Gemini | Langchain AI Agent (Google Gemini) | Generate full Markdown blog post | Summarize Search Results, Wikipedia Tool | Prepare File for Commit | ## Ai agent<br>Standardized prompt generating well-structured markdown blog post. |
|
||||
| Prepare File for Commit | Set | Prepare filename, content, commit message | Generate Blog with Gemini | Commit Blog Post to GitHub | ## Set the file for github upload<br>Prepares content for seamless GitHub integration. |
|
||||
| Commit Blog Post to GitHub | GitHub | Commit Markdown blog post to GitHub repo | Prepare File for Commit | Mark Topic as Done in Sheet | ## Github upload<br>Uploads the blog post to the GitHub repository. |
|
||||
| Mark Topic as Done in Sheet | Google Sheets | Update blog status to “done” in Google Sheet| Commit Blog Post to GitHub | None | ## Change status<br>Marks the blog topic status in the sheet as done after upload. |
|
||||
| Sticky Note | Sticky Note | Descriptive notes | None | None | ## Automated Blog Creation description and instructions. |
|
||||
| Sticky Note1 | Sticky Note | Trigger node comment | None | None | ## Trigger node configuration note. |
|
||||
| Sticky Note2 | Sticky Note | Google Sheets topic retrieval comment | None | None | ## Get Topic from spreadsheet comment. |
|
||||
| Sticky Note3 | Sticky Note | Extract topic comment | None | None | ## Extract topic comment. |
|
||||
| Sticky Note4 | Sticky Note | Deep search explanation | None | None | ## Deep Search about topic comment. |
|
||||
| Sticky Note5 | Sticky Note | Formatting search results explanation | None | None | ## Format the result comment. |
|
||||
| Sticky Note6 | Sticky Note | AI agent comment | None | None | ## Ai agent comment. |
|
||||
| Sticky Note7 | Sticky Note | File preparation comment | None | None | ## Set the file for github upload comment. |
|
||||
| Sticky Note8 | Sticky Note | GitHub upload comment | None | None | ## Github upload comment. |
|
||||
| Sticky Note9 | Sticky Note | Status update comment | None | None | ## Change status comment. |
|
||||
| Sticky Note10 | Sticky Note | Setup and integration instructions | None | None | # n8n Integrations & GitHub Pages Jekyll Setup with prerequisites and stepwise instructions. |
|
||||
|
||||
---
|
||||
|
||||
### 4. Reproducing the Workflow from Scratch
|
||||
|
||||
1. **Create a Schedule Trigger node**
|
||||
- Set trigger time to daily at 09:00 (adjust timezone as needed).
|
||||
- No credentials required.
|
||||
|
||||
2. **Add a Google Sheets node “Get Topic from Google Sheet”**
|
||||
- Operation: Read rows
|
||||
- Sheet: `gid=0` of your Google Sheet document (containing blog topics).
|
||||
- Filter: Only rows where `status` column is not “done”.
|
||||
- Connect input from Schedule Trigger.
|
||||
- Configure Google Sheets OAuth2 credentials.
|
||||
|
||||
3. **Add a Set node “Extract Blog Title”**
|
||||
- Assign a new variable `topic` set to the field containing the blog title from the sheet (e.g., `{{$json.Title}}`).
|
||||
- Connect input from “Get Topic from Google Sheet”.
|
||||
|
||||
4. **Add a Tavily Web Search node**
|
||||
- Configure query to use `{{$json.topic}}`.
|
||||
- Set search depth to “advanced” for richer results.
|
||||
- Connect input from “Extract Blog Title”.
|
||||
- Add Tavily API credentials.
|
||||
|
||||
5. **Add a Code node “Summarize Search Results”**
|
||||
- Use JS code to select top 3 Tavily results, format with title, snippet, and source URL into a Markdown context string.
|
||||
- Input from Tavily Web Search node.
|
||||
|
||||
6. **Add a Wikipedia Tool node**
|
||||
- Set to fetch Wikipedia content dynamically based on topic.
|
||||
- Connect input from Tavily node via AI tool interface or directly integrate with Langchain AI agent.
|
||||
|
||||
7. **Add a Langchain Agent node “Generate Blog with Gemini”**
|
||||
- Use Google Gemini 2.5 Pro model with your Google Palm API credentials.
|
||||
- Set prompt to generate a complete Markdown blog post with strict Jekyll front matter and formatting rules.
|
||||
- Inject variables `topic` and `context` from “Summarize Search Results” and Wikipedia content.
|
||||
- Connect inputs accordingly.
|
||||
|
||||
8. **Add a Set node “Prepare File for Commit”**
|
||||
- Create filename as `YYYY-MM-DD-topic-slug.md` where topic slug is lowercased and spaces replaced by hyphens.
|
||||
- Assign `fileContent` to the AI-generated Markdown output.
|
||||
- Compose a commit message like `Add blog post: [Topic Title]`.
|
||||
- Input from AI agent output.
|
||||
|
||||
9. **Add a GitHub node “Commit Blog Post to GitHub”**
|
||||
- Configure repository owner and name.
|
||||
- Set file path to `_posts/{{filePath}}`.
|
||||
- Use commit message and file content from previous node.
|
||||
- Use OAuth2 credentials for GitHub.
|
||||
- Connect input from “Prepare File for Commit”.
|
||||
|
||||
10. **Add a Google Sheets node “Mark Topic as Done in Sheet”**
|
||||
- Update operation on the same Google Sheet.
|
||||
- Match row by `row_number` from input.
|
||||
- Update `status` column to “done”.
|
||||
- Input from GitHub commit node.
|
||||
- Use Google Sheets OAuth2 credentials.
|
||||
|
||||
---
|
||||
|
||||
### 5. General Notes & Resources
|
||||
|
||||
| Note Content | Context or Link |
|
||||
|--------------------------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------|
|
||||
| Automated blog creation workflow that integrates LLM (Google Gemini), Tavily web search, Wikipedia, Google Sheets, and GitHub publishing. | Workflow description sticky note in the original workflow. |
|
||||
| Instructions for setting up Tavily API key: Sign up at https://www.tavily.com/, get API key from dashboard. | Sticky note with setup instructions for prerequisites. |
|
||||
| Google Sheets, GitHub OAuth2, and Google Gemini credentials must be configured in n8n with appropriate permissions. | Credential setup reminders from sticky notes and workflow nodes. |
|
||||
| Jekyll blog posts require Markdown files with YAML front matter in `_posts/` folder - filename format: `YYYY-MM-DD-title.md`. | Formatting rules embedded in AI prompt and GitHub commit node configuration. |
|
||||
| For GitHub Pages + Jekyll, ensure repository is configured correctly with Pages enabled and `_config.yml` set for your blog site. | Setup instructions in sticky notes and GitHub publishing node context. |
|
||||
| The workflow uses advanced prompt engineering to ensure the blog post is ready to publish without further manual editing. | Prompt details in the Gemini AI node configuration. |
|
||||
|
||||
---
|
||||
|
||||
This documentation enables advanced users and automation agents to understand, reproduce, and maintain the “Automated Blog Generation with Gemini AI, GitHub & Jekyll Publishing” workflow in n8n. It covers all nodes, logical flow, integration points, and potential failure scenarios.
|
||||
Reference in New Issue
Block a user