Content Automation MCP Server
Enables automated content generation and scheduling for Instagram and Pinterest via the Late API. It allows users to manage publications, generate AI-powered captions, and schedule posts or stories using natural language commands.
README
Content Automation
Automated content publishing for Pinterest and Instagram with AI-powered content generation and scheduling.
Features
- ā
MCP Integration ā - Natural language control through AI assistants (primary way of working)
- š Detailed MCP description - setup, prompt examples, troubleshooting
- ā TypeScript - Full type safety for reliability and better IDE support
- ā Late API - Publishing to Instagram and Pinterest through a unified API
- ā Unique AI-powered content generation for Pinterest and Instagram (OpenAI with customizable prompts)
- ā Publication scheduling through Late API (no local scheduler needed)
- ā Instagram Stories Support - Publishing Stories through Late API
- ā Publishing to both platforms with a single request
- ā Processed image tracking
- ā OpenAI integration for content generation (optional)
- ā CLI commands - alternative way of working through terminal
Installation
- Install dependencies:
npm install
- Build the TypeScript project:
npm run build
- Copy
.env.exampleto.envand fill it:
cp .env.example .env
- Configure environment variables in
.env:- Late API (required): Get API key at getlate.dev
- CDN Base URL (required): Specify the base CDN URL where images are hosted
- OpenAI API (optional): For AI-powered content generation
- Content Prompt (optional): Customize content generation by copying
prompts/content-prompt.example.txttoprompts/content-prompt.txtand editing it
Getting API Keys
Late API (Required) ā
The project uses only Late API for publishing to Instagram and Pinterest.
- Register at getlate.dev
- Get API key
- Connect Instagram and Pinterest accounts
- Get Profile ID and Account IDs from the dashboard
- Add to
.env:LATE_API_KEY=your_key LATE_PROFILE_ID=your_profile_id LATE_INSTAGRAM_ACCOUNT_ID=your_instagram_account_id LATE_PINTEREST_ACCOUNT_ID=your_pinterest_account_id
Late API Benefits:
- ā No local scheduler needed - Late API publishes on schedule automatically
- ā Reliable scheduling on Late servers
- ā Publishing to Instagram AND Pinterest with a single request - saves API requests!
- ā Free plan available (60 requests/minute)
- ā Support for 13 platforms (Instagram, Pinterest, Facebook, LinkedIn, Twitter/X, TikTok, YouTube, Reddit, Bluesky, Threads, Google Business, Telegram, Snapchat)
Supported Platforms: Instagram, Pinterest, Facebook, LinkedIn, Twitter/X, TikTok, YouTube, Reddit, Bluesky, Threads, Google Business, Telegram, Snapchat
Usage
The project supports two ways of working:
- Through MCP agent (recommended) - Natural language control through AI assistant
- Through CLI commands - Traditional way through terminal
Setup Check
Before starting, check API connection:
npm run check
This will check:
- Late API connection (required)
- CDN configuration (required)
- OpenAI API availability (optional)
CLI Commands
Publishing Images
npm run publish -- --image=IMG_5857.jpg [options]
Important:
- Images must be on CDN. Specify only the filename (e.g.,
IMG_5857.jpg), the system will automatically build the CDN URL. - Case sensitivity: Filename is case-sensitive. Specify the filename exactly as it appears on CDN (e.g.,
IMG_5857.jpg, notimg_5857.jpg).
Examples:
# Default: all three types (Instagram post, Instagram story, Pinterest pin)
npm run publish -- --image=IMG_5857.jpg
# Publish in 1 hour
npm run publish -- --image=IMG_5857.jpg 1
# Draft with publication date
npm run publish -- --image=IMG_5857.jpg --date="tomorrow 18:00" --draft
# With context
npm run publish -- --image=IMG_5857.jpg 1 --context="City center"
# Instagram post only
npm run publish -- --image=IMG_5857.jpg --post
# Instagram story only
npm run publish -- --image=IMG_5857.jpg --story
# Instagram story + Pinterest pin (comma-separated)
npm run publish -- --image=IMG_5857.jpg --types=story,pin
# Instagram post + Pinterest pin
npm run publish -- --image=IMG_5857.jpg --types=post,pin
# Pinterest pin only
npm run publish -- --image=IMG_5857.jpg --pin
# With draft
npm run publish -- --image=IMG_5857.jpg --story --draft
Options:
--image=<filename>- Image filename (required in CLI mode)--context=<text>- Context for content generation--types=<list>- Content types comma-separated:post,story,pin(e.g.,--types=story,pin)--post- Instagram post only (short form for--types=post)--story- Instagram story only (short form for--types=story)--pin- Pinterest pin only (short form for--types=pin)- (no flags) - All three types (default): Instagram post, Instagram story, Pinterest pin
--date=<date>- Publication date (YYYY-MM-DD HH:MM or "tomorrow 18:00")<hours>- Alternative to --date: number of hours from current moment--draftor-d- Create as draft
Content Types:
- Default (nothing specified) - Publishes all three types: Instagram post, Instagram story, Pinterest pin
--types=post,story,pin- Specify types comma-separated (e.g.,--types=story,pin)--post- Publishes only Instagram post--story- Publishes only Instagram story--pin- Publishes only Pinterest pin (without Instagram)
Detailed Documentation:
- CLI.md - Detailed CLI description with usage examples
Building the Project
# Build TypeScript to JavaScript
npm run build
# Build and watch for changes (auto-recompiles on file changes)
npm run watch
Using MCP in Dev Mode
When developing with MCP, use automatic recompilation:
# Terminal: Run watch for automatic recompilation
npm run watch
How it works:
watchautomatically recompiles TypeScript when files change- Cursor automatically restarts MCP server when
dist/mcp/mcp-server.jschanges - You can immediately use updated MCP tools in Cursor chat
Important: After changing code in src/mcp/ or related modules, changes will apply automatically through watch + MCP restart in Cursor.
š§ Dev Mode (dry-run):
- In
watchmode,DEV_MODE=trueis automatically enabled - Requests to Late API are NOT sent - instead, JSON ready to send is output
- This allows testing logic without real publications
- For real requests, use
npm run watch:prodor simplynpm run build+ commands
Image Processing
Publishing a Single Image
# With specific image
npm run publish -- --image=IMG_0802.JPG
# With context
npm run publish -- --image=IMG_0802.JPG --context="Concert hall, evening"
# With content type
npm run publish -- --image=IMG_0802.JPG --context="City center" --story
# Instagram post only
npm run publish -- --image=IMG_0802.JPG --context="City center" --post
# With publication date
npm run publish -- --image=IMG_0802.JPG --context="Urban landscape" --date="2025-01-25 18:00"
# All parameters together
npm run publish -- --image=IMG_0802.JPG --context="Concert hall, evening" --post --date="2025-01-25 18:00"
Note: Local scheduler is not needed - Late API automatically publishes posts on schedule.
Using MCP Agent (Recommended) ā
MCP (Model Context Protocol) - This is the primary and recommended way to work with the project. Manage publications through AI assistant using natural language directly in Cursor, Claude, or other AI tools.
Quick Start with MCP
-
Build the project:
npm run build -
MCP server is automatically available in Cursor (via
.cursor/mcp.json) -
Use natural language:
"Publish IMG_0802.JPG tomorrow at 18:00 with context 'Concert hall, evening'" "Show list of available images" "Generate content for IMG_1460.JPG" "Schedule Story for IMG_2228.JPG today at 20:00"
MCP Benefits:
- ā Natural language - No need to remember commands
- ā AI Integration - Works directly in Cursor/Claude
- ā Smart automation - AI can plan publications itself
- ā Contextual understanding - AI understands your intentions
Detailed Documentation:
- MCP.md - Setup, prompt examples, troubleshooting
Using CLI Commands
Alternative way of working through terminal. Detailed documentation: CLI.md
Project Structure
The project is organized on a modular principle for better scalability and support.
Main Modules:
src/services/- API services (Late API, Pinterest API)src/generators/- Content generators (AI-powered with customizable prompts)src/processors/- Data processors (images)src/utils/- Utilities (CDN, configuration, scheduling)src/cli/- CLI commands and entry pointssrc/mcp/- MCP integration for AI assistantssrc/types/- TypeScript types and interfacesprompts/- Content generation prompts (customizable)
Modular Structure Benefits:
ā
Separation of concerns - Each module is responsible for its area
ā
Singleton pattern - Services are exported as singleton instances
ā
Direct imports - Simplified structure without intermediate index.ts
ā
Scalability - Easy to add new services
ā
Testability - Each module can be tested separately
Content Format
- Title: SEO-optimized headline
- Description: Description with keywords
- Keywords: List of keywords
- CTA: Call to action (leads to Instagram)
Instagram Posts
- Caption: Emotional story text
- Soft CTA: Soft call to action
- Hashtags: 3-5 relevant hashtags (customizable via content prompt)
Instagram Stories
- Stories do not support long captions
- Text overlays can be used (via API)
- User mentions (user_tags) are supported
- Automatically disappear after 24 hours
Limitations and Capabilities
Late API
- ā Only Late API is used for publishing to Instagram and Pinterest
- ā Native scheduling through Late API
- ā Does not require local scheduler
- ā Publishing to both platforms with a single request
- ā Support for media file upload or using public CDN URLs
- ā Supports Stories - Can publish Stories through Late API
- ā ļø Stories limitations: Can only schedule for the next 24 hours
- ā ļø Free plan: 60 requests/minute
CDN for Images
Images must be hosted on a public CDN:
- Specify
CDN_BASE_URLin.env(required) - Images are accessed directly from CDN (not uploaded to Late API)
- Specify only the filename in commands (e.g.,
IMG_5857.jpg) - The system automatically builds CDN URLs by content type:
postā{CDN_BASE_URL}/post/{filename}storyā{CDN_BASE_URL}/story/{filename}pinā{CDN_BASE_URL}/pin/{filename}
- Important: Filenames are case-sensitive - use exact case as on CDN
Content Generation Customization
The content generation prompt can be customized to match your brand voice and target audience:
-
Copy the example prompt:
cp prompts/content-prompt.example.txt prompts/content-prompt.txt -
Edit
prompts/content-prompt.txtwith your custom instructions:- Adjust tone and style
- Modify hashtag requirements
- Change content focus and goals
- Update output format if needed
-
The custom prompt is automatically loaded - no code changes needed!
Note: prompts/content-prompt.txt is ignored by git, so your customizations remain private. The example file (content-prompt.example.txt) serves as a template and is version-controlled.
Security
- Never commit
.envfile - Never commit
prompts/content-prompt.txt(already in.gitignore) - Store API keys in a secure place
- Regularly check Late API key expiration
- Do not share API keys publicly
Troubleshooting
Late API Connection Error
- Check API key correctness:
npm run check - Make sure both accounts (Instagram and Pinterest) are connected in Late
- Check API key expiration
Publication Error
- Make sure Instagram and Pinterest accounts are connected in Late
- Check account access rights
- Make sure images are available on CDN at specified URLs
Images Not Processing
- Check file formats (supported: JPG, PNG)
- Make sure files are not corrupted
- Check filename case - Filename is case-sensitive (e.g.,
IMG_5857.jpgāimg_5857.jpg) - Make sure file exists on CDN at specified path
CDN Not Working
- Make sure
CDN_BASE_URLis specified correctly - Check that images are available at specified URLs
- Make sure URLs are publicly accessible
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.