Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@self-service-mcp look up Jane Doe in the company directory"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A workplace self-service MCP server: directory, time off, tickets, policy search. Writes are governed. Compensation is structurally out of scope — the tool exists so the model can be told no instead of hallucinating a yes.
The interesting part of enterprise agents is not “can it call an API.” It is which APIs it must never call, even if the user (or a prompt injection) asks nicely.
search_directory read
get_time_off_balance read
search_policy read
submit_time_off write → shadow / approve
create_ticket write → shadow / approve
change_compensation always DENYSynthetic directory data only. No employer systems, no personal data from work.
npm install
npm test
npm run mcpRelated MCP server: HR System MCP Server
Same choke-point as lattice-mcp and ewm-ops-mcp.
Default SS_MODE=shadow.
change_compensation is registered on purpose. Hidden capabilities are how agents surprise you. Denied capabilities are how you sleep.
MIT © Prasanna Kumar Popuri
This server cannot be deployed
Read-only finance and operations controls for AI agents with evidence and safe next actions.
Governed HR records and confirmed time-off actions for UK SMEs.
Read-only employment-law research: 140K+ rulings, employer track records, rights & EEOC guides.
Give AI agents 30,000+ safe, token-optimized actions across Workday, SAP, Oracle + hundreds more.
