Hyperpost
An AI-native publishing engine for persona-driven content creation and multi-platform publishing.
Hyperpost
An AI-native publishing engine built on Model Context Protocol (MCP). Hyperpost enables persona-driven content creation and multi-platform publishing through a unified AI-first interface.
Overview
Hyperpost is designed from the ground up to work with AI agents like Claude. It uses the Model Context Protocol to expose all functionality, making it equally accessible to AI assistants, command-line tools, and web interfaces.
Key Features
- MCP-First Architecture: All operations exposed through Model Context Protocol
- Persona System: Create distinct voices for different contexts (personal blog, work, research)
- Multi-Platform Publishing: Write once, publish to blog, LinkedIn, Twitter/X
- AI-Native: Designed for AI agents to create, refine, and manage content
- Zero Dependencies: Follows hyperserve's philosophy of minimal external dependencies
Quick Start
For Claude Desktop Users
-
Build the MCP server:
make build-mcp -
Add to Claude Desktop configuration:
{ "mcpServers": { "hyperpost": { "command": "/path/to/hyperpost-mcp", "args": ["-content", "/path/to/content"] } } } -
Restart Claude Desktop and start creating content!
For Developers
# Clone the repository
git clone https://github.com/osauer/hyperpost
cd hyperpost
# Build everything
make build-mcp # MCP server
make build # CLI client (optional)
# Run MCP server
./hyperpost-mcp -content ./content
# Or use the CLI
./hyperpost persona create -id tech-writer -name "Tech Writer" -bio "Technical content creator"
./hyperpost content create -persona tech-writer -title "My First Post"
Architecture
Hyperpost uses an MCP-first architecture where all functionality is exposed through the Model Context Protocol:
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Claude Desktop │ │ CLI Client │ │ Web UI │
│ (MCP Client) │ │ (MCP Client) │ │ (MCP Client) │
└────────┬────────┘ └────────┬─────────┘ └────────┬────────┘
│ │ │
└───────────────────────┴─────────────────────────┘
│
┌────────────▼────────────┐
│ MCP Server (stdio) │
│ hyperpost-mcp │
└────────────┬────────────┘
│
┌────────────▼────────────┐
│ Core Services │
│ (personas, content, │
│ render, publish) │
└─────────────────────────┘
MCP Tools Available
Persona Management
persona_create- Create a new persona with unique voice and contextpersona_list- List all configured personaspersona_get- Get detailed information about a personapersona_delete- Remove a persona
Content Management
content_create- Create new content with markdown and metadatacontent_list- List content with optional filters (persona, draft status)content_get- Retrieve specific contentcontent_update- Update existing contentcontent_delete- Remove content
AI Tools
ai_generate- Generate content using persona contextai_refine- Refine existing content to match persona voice
Project Structure
hyperpost/
├── cmd/
│ ├── hyperpost-mcp/ # MCP server (primary interface)
│ └── hyperpost/ # CLI client
├── internal/ # Core implementation
│ ├── mcp/ # MCP protocol server
│ ├── persona/ # Persona management
│ └── content/ # Content handling
└── content/ # Your content (git-tracked)
└── personas/ # Persona-specific content
Documentation
- CLAUDE.md - Instructions for AI agents working with the codebase
- ARCHITECTURE.md - Detailed architectural decisions
- MCP_SETUP.md - MCP server setup guide
- CLAUDE_USAGE_EXAMPLES.md - Examples of using with Claude
- API.md - Complete MCP tools documentation
Philosophy
Hyperpost embraces several core principles:
- AI-First: Designed for AI agents as primary users
- Zero Dependencies: Minimal external dependencies for security and simplicity
- Git-Native: All content is version-controlled
- Privacy-First: Your content stays local
- Extensible: Easy to add new platforms and features
Status
Hyperpost is in active development. Current status:
- ✅ MCP server implementation
- ✅ Persona management
- ✅ Content CRUD operations
- ✅ AI tool interfaces
- 🚧 Rendering engine
- 🚧 Social media publishing
- 🚧 Web interface
Contributing
When contributing, please:
- Follow the zero-dependency philosophy
- Ensure all features are exposed through MCP
- Update documentation for AI agents
- Add tests for new functionality
License
MIT License - see LICENSE file for details
Built With
- Go 1.24+
- Hyperserve - Lightweight web server
- Model Context Protocol - AI-native interface protocol
相關伺服器
AI Collaboration MCP Server
An MCP server for AI-to-AI collaboration, enabling autonomous workflows and role-based task management between different AI models.
VNC
Remotely control any system running a VNC server, including Windows, Linux, and macOS, through an AI agent.
Kit.com (formerly ConvertKit) MCP
Manage your email lists, subscribers, broadcasts, sequences, and more through natural language.
Basecamp by CData
A read-only MCP server by CData that allows LLMs to query live Basecamp data.
GSuite
Interact with Google products, including Gmail and Calendar.
MCP Google Workspace
A comprehensive MCP server for managing Google Workspace services like Calendar, Contacts, and Gmail using OAuth2 authentication.
PowerPoint Translator
Translate PowerPoint files using AWS Bedrock. Requires AWS credentials to be configured.
AI Humanize MCP Server
Refines AI-generated content to sound more natural and human-like using advanced text enhancement.
mcp-server-nano-banana
AI image generation with Google Gemini's Nano-Banana model.
Vynn
Self-improving AI workflows with natural language backtesting. 21 MCP tools for creating workflows, backtesting trading strategies, parameter sweeps, portfolio optimization, prompt optimization, cron scheduling, and webhook triggers. Install: pip install vynn-mcp