lab/nusquama-n8nworkflows-xyz

Fork 0

mirror of https://github.com/nusquama/n8nworkflows.xyz.git synced 2025-11-27 01:35:37 +08:00

Files

nusquama 68df6e9c2f creation

2025-11-25 10:38:12 +01:00

22 KiB

Raw Blame History

Generate LinkedIn Activity Reports via Slack Commands with GPT-4.1 and Email

https://n8nworkflows.xyz/workflows/generate-linkedin-activity-reports-via-slack-commands-with-gpt-4-1-and-email-11052

Generate LinkedIn Activity Reports via Slack Commands with GPT-4.1 and Email

1. Workflow Overview

This n8n workflow enables users to generate detailed LinkedIn activity reports by invoking a Slack slash command. When a user types /check-linkedin [firstName lastName] in Slack, the workflow extracts the name, finds the corresponding LinkedIn profile using Apify actors, scrapes recent posts, and uses GPT-4.1 AI models to analyze posting frequency, topics, and highlights. Finally, it composes a formatted HTML report and sends it via email to a designated recipient.

The workflow is logically divided into these blocks:

1.1 Slack Command Intake & Validation: Receive Slack slash command input, validate if it contains a full name.
1.2 Name Extraction: Use GPT to extract first and last names from the input.
1.3 LinkedIn Profile Lookup: Use Apify actors to find the LinkedIn profile URL for the extracted name.
1.4 LinkedIn Posts Scraping: Scrape recent LinkedIn posts of the profile.
1.5 Posts Aggregation: Aggregate and structure the scraped posts for analysis.
1.6 AI Analysis & Summarization: Use GPT-4.1 and a Langchain structured output parser to analyze posts and generate a detailed summary.
1.7 HTML Report Generation & Email Delivery: Format the AI summary into an HTML report and email it to the requester.
1.8 Error Handling & Slack Feedback: Notify Slack if the input name is invalid.

2. Block-by-Block Analysis

1.1 Slack Command Intake & Validation

Overview:
This block receives the Slack slash command POST request, checks if the input text is a full name, branching to error messaging if not.

Nodes Involved:

Webhook
GPT 4.1-mini for classification
Do we have a full name ? (text classifier)
Tell on slack that no full name was found

Node Details:

Webhook
- Type: HTTP Webhook (POST)
- Configuration: Receives Slack slash command at a defined path (placeholder to be replaced)
- Input: Slack slash command payload (JSON)
- Output: Passes the raw Slack command data downstream
- Edge cases: Invalid HTTP method or malformed requests
GPT 4.1-mini for classification
- Type: OpenAI GPT chat model (GPT 4.1-mini)
- Role: Classify if input text is a full name or not
- Input: {{ $json.body.text }} (text entered in Slack command)
- Output: Prediction used for branching
- Edge cases: API errors, rate limiting, classification uncertainty
Do we have a full name ?
- Type: Text Classifier node (Langchain)
- Role: Categorize input as "full-name" or "not-full-name"
- InputText: {{ $json.body.text }}
- Categories: "full-name", "not-full-name"
- Output: Branches workflow based on category
- Edge cases: Misclassification, expression evaluation failures
Tell on slack that no full name was found
- Type: Slack node (chat message)
- Role: Sends a message back to the Slack channel notifying the user that the input was not a full name
- Configuration: Uses OAuth2 Slack credential, posts to the originating channel {{ $json.body.channel_id }}
- Input: Triggered if name classification is negative
- Edge cases: Slack API auth failures, invalid channel IDs

1.2 Name Extraction

Overview:
Extract first name and last name explicitly from the validated input text using GPT.

Nodes Involved:

GPT 4.1-mini to extract firstName + lastName
Extract firstName and lastName (Information Extractor node)

Node Details:

GPT 4.1-mini to extract firstName + lastName
- Type: OpenAI GPT chat model (GPT 4.1-mini)
- Role: Refine and normalize the name extraction process
- Input: Raw Slack command text
- Output: Passes text to the information extractor node
- Edge cases: API errors, ambiguous names
Extract firstName and lastName
- Type: Langchain Information Extractor
- Role: Identifies and extracts structured attributes: firstName and lastName
- Input: Text from previous GPT node
- Output: JSON with firstName and lastName fields (both required)
- Edge cases: Missing attributes, parsing failures

1.3 LinkedIn Profile Lookup

Overview:
Find the LinkedIn profile URL for the extracted person using Apify actor.

Nodes Involved:

Find Linkedin Profile (Apify actor)
Get Linkedin Profile URL (Apify dataset query)

Node Details:

Find Linkedin Profile
- Type: Apify actor call
- Role: Runs an Apify actor (configured via URL) that searches LinkedIn based on the first and last name
- Input: JSON body with firstName and lastName
- Output: Returns datasetId for collected profile data
- Edge cases: Actor timeout, invalid input, no profile found
Get Linkedin Profile URL
- Type: Apify dataset query
- Role: Retrieves profile URL from Apify dataset using the datasetId from previous node
- Input: Uses defaultDatasetId from previous node
- Output: Extracted LinkedIn profile URL (linkedinProfileUrl)
- Edge cases: Dataset empty, API errors

1.4 LinkedIn Posts Scraping

Overview:
Scrape recent posts of the found LinkedIn profile via another Apify actor.

Nodes Involved:

Scrap what this person posted recently (Apify actor)
Structure recent posts (Apify dataset query)
Aggregate (Aggregate node)

Node Details:

Scrap what this person posted recently
- Type: Apify actor call
- Role: Runs Apify actor to scrape LinkedIn posts for the given profile URL
- Input: JSON body with usernames array containing the LinkedIn profile URL
- Output: Dataset id for scraped posts
- Edge cases: Actor failure, rate limits, empty posts
Structure recent posts
- Type: Apify dataset query
- Role: Retrieves up to 20 recent posts from Apify dataset
- Input: Dataset ID from scraping node
- Output: List of post objects with text and posted_at.date properties
- Edge cases: Empty dataset, partial data
Aggregate
- Type: Aggregate node (standard n8n)
- Role: Collects and merges fields posted_at.date and text from the posts into a single object
- Input: List of recent posts
- Output: Aggregated JSON data for analysis
- Edge cases: Empty input array, aggregation errors

1.5 AI Analysis & Summarization

Overview:
Analyze aggregated LinkedIn posts data with GPT 4.1 to generate a detailed analytical summary following a strict JSON schema.

Nodes Involved:

Summer McBriefing (Langchain Agent)
Structured Output Parser (Langchain Output Parser)
GPT 4.1 (OpenAI chat model)

Node Details:

Summer McBriefing
- Type: Langchain Agent (AI language model with custom prompt)
- Role: Receives aggregated post dates and text, generates a JSON report including posting frequency, topics, highlights, and data quality stats
- Prompt: Detailed instructions to produce exactly the required JSON schema, no extra fields or prose
- Input: Aggregated dates and texts from previous nodes
- Output: JSON object matching LinkedInPostsAnalysis schema
- Edge cases: Parsing ambiguity, incomplete input data, GPT API errors
Structured Output Parser
- Type: Langchain Output Parser
- Role: Validates and enforces JSON structure according to the provided JSON schema
- Input: Raw JSON from Summer McBriefing
- Output: Structured, validated JSON for further processing
- Edge cases: Schema validation failures, parser errors
GPT 4.1
- Type: OpenAI GPT chat model (GPT 4.1)
- Role: Connected as AI languageModel input to Summer McBriefing (used internally)
- Edge cases: API failures, rate limits

1.6 HTML Report Generation & Email Delivery

Overview:
Generate an HTML report from the structured AI output and send it by email to a configured recipient.

Nodes Involved:

HTML (HTML generation)
Send report via Email (Gmail node)
Sticky Note12 (Contact info)

Node Details:

HTML
- Type: HTML node (n8n built-in)
- Role: Formats the JSON report into a styled, professional HTML email
- Uses Mustache-style expressions to inject AI output and extracted names into the template
- Includes summary, posting frequency, topics, and highlights in a visually clear layout
- Edge cases: Missing JSON fields, malformed HTML
Send report via Email
- Type: Gmail node (OAuth2)
- Role: Sends the generated HTML email to a fixed recipient (admin@example.com)
- Subject includes person’s name dynamically: Recent LinkedIn Activity about [personName]
- Requires Gmail OAuth2 credentials
- Edge cases: Email sending failures, auth errors, invalid recipient
Sticky Note12
- Type: Sticky Note
- Role: Contains contact information for questions or feedback (email and LinkedIn)
- No direct functional role in automation

1.7 Error Handling & Slack Feedback

Overview:
If the input text is not a full name, this block provides user feedback on Slack.

Nodes Involved:

Tell on slack that no full name was found
Sticky Note1 (Error handling explanation)

Node Details:

Tell on slack that no full name was found
- As described in 1.1
Sticky Note1
- Provides a note reminding the user about error handling for no full name found input

3. Summary Table

Node Name	Node Type	Functional Role	Input Node(s)	Output Node(s)	Sticky Note
Webhook	Webhook	Receive Slack slash command request	—	GPT 4.1-mini for classification
GPT 4.1-mini for classification	OpenAI GPT chat model	Classify if input is a full name	Webhook	Do we have a full name ?
Do we have a full name ?	Text classifier (Langchain)	Branch workflow on full name check	GPT 4.1-mini for classification	Extract firstName and lastName / Tell on slack no full name	Sticky Note1: No full name found error handling
Tell on slack that no full name was found	Slack message node	Notify user on Slack of invalid input	Do we have a full name ?	—	Sticky Note1: No full name found error handling
GPT 4.1-mini to extract firstName + lastName	OpenAI GPT chat model	Extract first and last name from input	Do we have a full name ?	Extract firstName and lastName
Extract firstName and lastName	Langchain Information Extractor	Extract structured firstName and lastName	GPT 4.1-mini to extract names	Find Linkedin Profile
Find Linkedin Profile	Apify actor	Find LinkedIn profile URL from name	Extract firstName and lastName	Get Linkedin Profile URL	Sticky Note3: Data gathering via Apify
Get Linkedin Profile URL	Apify dataset query	Retrieve LinkedIn profile URL	Find Linkedin Profile	Scrap what this person posted recently	Sticky Note3: Data gathering via Apify
Scrap what this person posted recently	Apify actor	Scrape recent LinkedIn posts	Get Linkedin Profile URL	Structure recent posts	Sticky Note3: Data gathering via Apify
Structure recent posts	Apify dataset query	Retrieve structured posts (limit 20)	Scrap what this person posted recently	Aggregate	Sticky Note3: Data gathering via Apify
Aggregate	Aggregate node	Aggregate post dates and texts	Structure recent posts	Summer McBriefing	Sticky Note4: Summarization & Intelligence Layer
Summer McBriefing	Langchain Agent	Analyze posts and generate JSON report	Aggregate	HTML	Sticky Note4: Summarization & Intelligence Layer
Structured Output Parser	Langchain Output Parser	Validate and structure AI output JSON	Summer McBriefing	Summer McBriefing (loop back)
HTML	HTML node	Format JSON summary into styled HTML email	Summer McBriefing	Send report via Email	Sticky Note5: HTML Output & Email
Send report via Email	Gmail node	Email the report to recipient	HTML	—
Sticky Note12	Sticky Note	Contact info for questions and feedback	—	—
Sticky Note	Sticky Note	Workflow overview and instructions	—	—
Sticky Note2	Sticky Note	Slack intake and parsing explanation	—	—
Sticky Note3	Sticky Note	Notes on Apify data gathering	—	—	Sticky Note3: Data gathering via Apify
Sticky Note4	Sticky Note	Summarization and AI layer explanation	—	—	Sticky Note4: Summarization & Intelligence Layer
Sticky Note5	Sticky Note	HTML output and email explanation	—	—	Sticky Note5: HTML Output & Response via Email
Sticky Note9	Sticky Note	Video walkthrough link	—	—	# 🎙️ Video Walkthrough @youtube

4. Reproducing the Workflow from Scratch

Create Webhook Node
- Type: Webhook
- HTTP Method: POST
- Path: /path-placeholder-to-be-replaced-by-your-webhook (replace with your actual path)
- Purpose: Receive Slack slash command payloads
Add GPT 4.1-mini for Classification Node
- Type: OpenAI GPT chat model
- Model: gpt-4.1-mini
- Input: Text from Webhook node — {{$json.body.text}}
- Purpose: Classify if input is a full name or not
Add Text Classifier Node "Do we have a full name ?"
- Type: Langchain Text Classifier
- Input Text: {{$json.body.text}}
- Categories:
  - full-name (input is a full name)
  - not-full-name (otherwise)
- Connect from GPT classification node
- Purpose: Branch the workflow
Add Slack Node for Invalid Name Feedback
- Type: Slack
- Action: Send Message to Channel
- Channel: {{$json.body.channel_id}} (from Webhook)
- Text: "{{$json.body.text}} is not a full name. Please send a full name to get information."
- Authentication: OAuth2 (Slack credentials)
- Connect from classifier node on not-full-name branch
Add GPT 4.1-mini to Extract First and Last Name
- Model: gpt-4.1-mini
- Input: Slack command text {{$json.body.text}}
- Connect from classifier node on full-name branch
Add Information Extractor Node
- Type: Langchain Information Extractor
- Input Text: Output text from previous GPT node
- Attributes:
  - firstName (required)
  - lastName (required)
- Connect from GPT extraction node
Add Apify Actor Node "Find Linkedin Profile"
- Actor URL: https://console.apify.com/actors/FbqC9BRstFBddhUqj/input
- Input JSON Body:
```
{
  "firstName": "{{ $json.output.firstName }}",
  "lastName": "{{ $json.output.lastName }}"
}
```
- Connect from Information Extractor node
- Credentials: Apify API key
Add Apify Dataset Query Node "Get Linkedin Profile URL"
- Resource: Datasets
- Dataset ID: ={{ $json.defaultDatasetId }} (from previous node)
- Connect from "Find Linkedin Profile"
Add Apify Actor Node "Scrap what this person posted recently"
- Actor URL: https://console.apify.com/actors/r4oNX7IHlW4RQAjKP
- Input JSON Body:
```
{
  "usernames": ["{{ $json.linkedinProfileUrl }}"]
}
```
- Connect from "Get Linkedin Profile URL"
- Credentials: Apify API key
Add Apify Dataset Query Node "Structure recent posts"
- Resource: Datasets
- Dataset ID: ={{ $json.defaultDatasetId }} (from previous node)
- Limit: 20
- Connect from "Scrap what this person posted recently"
Add Aggregate Node
- Fields to Aggregate:
  - posted_at.date
  - text
- Connect from "Structure recent posts"
Add Langchain Agent Node "Summer McBriefing"
- Text Input Template:
```
Dates:
{{ $json.date }}

Posts: 
{{ $json.text }}
```
- System Message: Detailed analytical instructions to produce JSON summary with posting frequency, topics, highlights, and data quality
- Input: Connect from Aggregate node
- Built-in Tools: GPT 4.1 model (configured)
- Credentials: OpenAI API key
Add Langchain Output Parser Node "Structured Output Parser"
- Schema: Use the detailed LinkedInPostsAnalysis JSON schema embedded in the workflow
- Connect as AI output parser to "Summer McBriefing" node
Add HTML Node
- Paste the provided HTML template from the workflow, including placeholders for dynamic data injection using n8n expressions, e.g., {{ $json.output.latest_post_date }}
- Connect from "Summer McBriefing" node output
Add Gmail Node "Send report via Email"
- Send To: admin@example.com (replace with desired recipient)
- Subject: Recent LinkedIn Activity about {{ $('Get Linkedin Profile URL').item.json.personName }}
- Message: ={{ $json.html }} (HTML output from previous node)
- Credentials: Gmail OAuth2 credentials configured in n8n
- Connect from HTML node
Set up Slack App and Slash Command
- Create Slack App with /check-linkedin slash command
- Set Request URL to your n8n webhook URL path
- Ensure Slack OAuth2 credentials have chat:write and commands scopes
Configure Credentials
- Apify API key with access to required actors
- OpenAI API key with GPT 4.1 and GPT 4.1-mini
- Slack OAuth2 token with required scopes
- Gmail OAuth2 account for sending emails
Testing & Validation
- Test slash command with valid full names to verify end-to-end functionality
- Test with invalid inputs to verify error handling messages in Slack
- Monitor Apify actor usage and costs
- Monitor OpenAI API usage and costs

5. General Notes & Resources

Note Content	Context or Link
Slack Slash Command: Run `/check-linkedin [firstName lastName]` in Slack to get LinkedIn activity.	Sticky Note overview at the start of the workflow
Workflow requires two Apify actors for LinkedIn profile lookup and post scraping.	Sticky Note3: Data gathering via Apify
Summarization and intelligence layer uses GPT 4.1 with a detailed JSON schema for output format.	Sticky Note4: Summarization & Intelligence Layer
HTML email format includes profile name, activity stats, topics, and highlights.	Sticky Note5: HTML Output & Response via Email
Video walkthrough available on YouTube: `@[youtube](-4VwNYsOIPc)`	Sticky Note9: Video Walkthrough
Contact for feedback: Emir Belkahia, email: emir.belkahia@gmail.com, LinkedIn: linkedin.com/in/emirbelkahia	Sticky Note12
Cost considerations: Apify actors and GPT calls incur usage costs; optimize parameters accordingly.	Sticky Note3 and Sticky Note4

This document comprehensively covers the workflow logic, node configurations, integration points, and reproduction steps to facilitate thorough understanding, modification, and deployment by advanced users or automation agents.

22 KiB Raw Blame History Unescape Escape

Generate LinkedIn Activity Reports via Slack Commands with GPT-4.1 and Email

1. Workflow Overview

2. Block-by-Block Analysis

1.1 Slack Command Intake & Validation

1.2 Name Extraction

1.3 LinkedIn Profile Lookup

1.4 LinkedIn Posts Scraping

1.5 AI Analysis & Summarization

1.6 HTML Report Generation & Email Delivery

1.7 Error Handling & Slack Feedback

3. Summary Table

4. Reproducing the Workflow from Scratch

5. General Notes & Resources

22 KiB

Raw Blame History