mcp-factorial
Secure, privacy-focused MCP server for FactorialHR. Access employee directories & org structure in Claude - deliberately excludes payroll/financial data for enterprise security.
claude mcp add --transport stdio t4dhg-mcp-factorial npx -y @t4dhg/mcp-factorial
How to use
The MCP FactorialHR server provides a hierarchical set of tools to manage and interact with FactorialHR data via the MCP protocol. It exposes 14 category-based tool endpoints (e.g., factorial_employees, factorial_teams, factorial_time_off, factorial_projects, factorial_documents, etc.) each with a defined set of actions such as list, get, create, update, delete, and specialized actions like search, terminate, approve, or reject. The server is designed to minimize context usage while preserving full CRUD capabilities and enforces safety guardrails for high-risk operations. You can discover available categories and actions using the factorial_discover tool and then invoke the appropriate tool with the action parameter to perform operations across employees, teams, locations, payroll, documents, and more. This makes it suitable for powering AI assistants (like Claude) with structured, auditable access to enterprise data, while maintaining safety and governance through explicit confirmations for risky operations and audit logging for write actions.
How to install
Prerequisites:
- Node.js 18+ installed on your system
- npm (comes with Node.js)
- Install Node.js (if not already installed)
- macOS (with Homebrew): brew install node
- Windows: download and install from https://nodejs.org/
- Linux (example using apt): sudo apt-get update sudo apt-get install -y nodejs npm
- Install and run the MCP FactorialHR server via npx
- This pulls and runs the MCP server package without permanently installing it locally: npx -y @t4dhg/mcp-factorial
- Verify installation
- After running, check the logs for startup messages indicating the MCP server is listening and ready to accept MCP requests.
- Optional: Persist installation locally (alternative)
- Install the package locally and run using node if a startup script is provided by the package:
mkdir mcp-factorial && cd mcp-factorial
npm init -y
npm install @t4dhg/mcp-factorial
Run the provided startup script (if available), e.g.:
npm run start
Additional notes
Tips and caveats:
- The server uses a hierarchical tool model with 14 categories; use factorial_discover to retrieve available categories and actions.
- For some operations, especially document downloads, OAuth2 may be required or enforced by the resource. Ensure appropriate authentication scopes.
- Enable safety guardrails for high-risk actions (terminate, delete) and review audit logs for all write operations.
- If you expose the MCP server to the internet, consider configuring proper authentication, rate limiting, and IP allowlists.
- Check for updates to the MCP FactorialHR package regularly to benefit from new categories, actions, and security improvements.
Related MCP Servers
mcp-framework
A framework for writing MCP (Model Context Protocol) servers in Typescript
docmole
Dig through any documentation with AI - MCP server for Claude, Cursor, and other AI assistants
obsidian
MCP server for Obsidian vault management - enables Claude and other AI assistants to read, write, search, and organize your notes
harvest
MCP server for Harvest time tracking API - use with Claude Desktop and other AI assistants
mcp-discovery
"MCP Discovery - The World's Largest MCP Server Index 14,000+ servers | Semantic search API | Real-time discovery Live API: https://mcp-discovery-two.vercel.app"
deliberate-reasoning-engine
MCP server that transforms linear AI reasoning into structured, auditable thought graphs