HoxCore
Enables LLMs to manage organizational metadata (projects, missions, activities, programs) via a file-based registry, providing CRUD operations and search capabilities through the Model Context Protocol.
README
HoxCore
<p align="center"> <img src="https://raw.githubusercontent.com/SDEscobedo/hoxcore/refs/heads/master/images/hoxcore_logo.jpg" alt="Hoxcore" width="400"> </p>
What is HoxCore?
HoxCore is a meta-manager — a low-level tool that centralizes the core metadata of organizational objects into a single, unified registry. Rather than managing execution or visualization directly, HoxCore acts as the foundational layer that independent software can build upon.
Organizational Object Types
HoxCore handles four categories of objects:
| Type | Orientation | Description |
|---|---|---|
| Projects | Goal-oriented | Has a defined finalization point; decoupled from execution |
| Missions | Event-oriented | Linked to execution; tied to a specific occurrence or execution window |
| Activities | Action-oriented | No definite end; represents indefinite, ongoing progression |
| Programs | Container | Groups and organizes Projects, Missions, and/or Activities |
Design Philosophy
HoxCore is intentionally minimal and low-level. It owns the metadata — everything else is up to you. Independent tools can be built on top of HoxCore for visualization, reporting, dashboards, or richer management interfaces, all reading from the same single source of truth.
Installation
From PyPI
pip install hoxcore
For Development
# Clone the repository
git clone https://github.com/SDEscobedo/hoxcore
cd hoxcore
# Create a virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install in development mode
pip install -e .
Usage
HoxCore manages your projects using a simple file-based system. It operates on a Registry (a directory on your disk) that holds various Entities (Programs, Projects, Missions, or Actions) stored as individual files. This mental model allows you to maintain full control over your data while using simple CLI commands to organize your workflow.
Available Commands
| Command | Description |
|---|---|
init |
Initialize a new registry |
create |
Create a new entity (project, program, mission, action) |
list |
List entities with optional filters |
show |
Show full details of a specific entity |
get |
Get a specific property from an entity |
edit |
Edit properties of an entity |
delete |
Delete an entity from the registry |
validate |
Validate registry integrity |
registry |
Manage registry location |
mcp |
Start the MCP server for LLM access |
Quick Start Guide
1. Initialize a Registry
Set up your first registry to start storing entities.
# Initialize a registry in the current directory
hxc init
# Initialize in a specific directory
hxc init ~/my-registry
# Initialize without git integration
hxc init --no-git
2. Create Entities
Add projects, programs, missions, or actions to your registry.
# Create a basic project
hxc create project "My First Project"
# Create a project with a description, tags, and due date
hxc create project "API Redesign" --description "Redesign the public API" --tags backend api --due-date 2025-06-01
# Create a program (container for projects)
hxc create program "Q2 Initiatives"
# Create a mission (execution-oriented, time-bound)
hxc create mission "Deploy v2.0"
# Create an action (ongoing, no end date)
hxc create action "Monitor system health"
3. List and Inspect
View your registry content and drill down into details.
# List all projects
hxc list project
# List all entities in the registry
hxc list all
# Filter by status or tag
hxc list project --status active
hxc list project --tag backend
# Show full details of a specific entity
hxc show <uid>
# Get a specific property
hxc get <uid> status
4. Edit and Delete
Update or remove entities as your tasks progress.
# Change status or add a tag
hxc edit <uid> --set-status completed
hxc edit <uid> --add-tag backend
# Set a new due date
hxc edit <uid> --set-due-date 2025-06-01
# Delete an entity (prompts for confirmation)
hxc delete <uid>
# Delete immediately without prompt
hxc delete <uid> --force
# Prevent automatic commit after deletion
hxc delete <uid> --no-commit
Getting Help
If you need more information about a specific command or general flags:
# General help
hxc --help
# Command-specific help
hxc create --help
hxc list --help
MCP Server (Model Context Protocol)
HoxCore includes a built-in MCP server that exposes registry functionality to LLMs through a standardized interface. This allows AI assistants (like Claude) to interact with your HoxCore registries directly.
Starting the Server
# Start with the default or configured registry
hxc mcp
# Start with a specific registry path
hxc mcp --registry /path/to/your/registry
# Specify transport (currently only stdio is supported)
hxc mcp --transport stdio
# Show server capabilities (tools, resources, prompts) without starting
hxc mcp --capabilities
# Start in read-only mode — write tools are not exposed to the LLM
hxc mcp --read-only
You can also start the server programmatically:
from hxc.mcp.server import create_server
# Normal mode (all tools)
server = create_server(registry_path="/path/to/registry")
server.run_stdio()
# Read-only mode (write tools omitted)
server = create_server(registry_path="/path/to/registry", read_only=True)
server.run_stdio()
Connecting to Claude (or other MCP-compatible clients)
Add HoxCore to your MCP client configuration. For Claude Desktop, update your claude_desktop_config.json:
{
"mcpServers": {
"hoxcore": {
"command": "hxc",
"args": ["mcp", "--registry", "/path/to/your/registry"]
}
}
}
To restrict the assistant to read-only access, add "--read-only" to the args list:
{
"mcpServers": {
"hoxcore": {
"command": "hxc",
"args": ["mcp", "--registry", "/path/to/your/registry", "--read-only"]
}
}
}
Available Tools
The MCP server exposes tools that an LLM can call. By default all seven tools are available; start with --read-only to expose only the four read tools.
| Tool | Type | Description |
|---|---|---|
list_entities |
Read | List entities in the registry, with optional filters for type, status, tags, category, and parent |
get_entity |
Read | Retrieve a specific entity by its ID or UID |
search_entities |
Read | Full-text search across entity titles and descriptions |
get_entity_property |
Read | Fetch a specific property from an entity, with support for list indexing and key filtering |
create_entity |
Write | Create a new entity (program, project, mission, or action) in the registry |
edit_entity |
Write | Update scalar fields or incrementally add/remove tags on an existing entity |
delete_entity |
Write | Delete an entity; requires a two-step confirmation (force=false first, then force=true) |
Available Resources
Resources are accessible via hxc:// URIs:
| URI | Description |
|---|---|
hxc://entity/{identifier} |
A specific entity by ID or UID (YAML) |
hxc://entities/{type} |
All entities of a given type (JSON) |
hxc://hierarchy/{identifier} |
Entity hierarchy and relationships (JSON) |
hxc://registry/stats |
Registry statistics and overview (JSON) |
hxc://search?q={query} |
Search results for a query (JSON) |
Extending the Server
You can register custom tools, resources, and prompts at runtime:
from hxc.mcp.server import create_server
server = create_server()
# Register a custom tool
def my_tool(registry_path=None, **kwargs):
"""My custom tool description."""
return {"result": "..."}
server.register_tool("my_tool", my_tool)
# Register a custom prompt
server.register_prompt({
"name": "my_prompt",
"description": "A helpful prompt template",
"arguments": [
{"name": "context", "description": "Context for the prompt", "required": True}
]
})
server.run_stdio()
Contributing
Contributions are welcome! Please see CONTRIBUTING for details on our development workflow, including branch naming and test requirements.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。