Skip to main content
Glama
Sign In
Telco Agent Lab

Telco Agent Lab

by mteinum
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Telco Agent Lab

A small, fully synthetic telecom sandbox for exploring AI agents and the Model Context Protocol (MCP). It exposes subscriber, network-registration and authentication-status tools, plus guarded demo operations for service provisioning and line suspension.

The project is designed for local experiments, demos and agent-evaluation suites. It has no carrier integrations, external services, API keys, subscriber data, authentication secrets or cryptographic vectors.

Why this exists

Telecom workflows combine structured operational data, sensitive customer context and actions with real-world consequences. This lab makes those boundaries tangible: customer case notes are explicitly marked untrusted, authentication tools expose status metadata only, and write tools require an explicit confirmation string and emit audit events.

The confirmation mechanism is a demo guard, not production authorization. A real deployment must obtain approval through an independent, authenticated control plane and enforce identity, authorization, idempotency, rate limits and durable auditing outside the model.

Related MCP server: ParcelPilot MCP Server

Run it

Requires Python 3.11+ and uv.

uv sync
uv run telco-agent-lab

The server speaks MCP over stdio. Configure your MCP-compatible client to launch uv run telco-agent-lab from this repository. The simulator starts fresh each time.

Tools

Tool

Purpose

list_subscribers

List synthetic IDs, plans and status

get_subscriber

Read a profile; case notes are labeled untrusted

get_network_status

Read synthetic registration metadata

get_auth_status

Read profile state without returning credentials or vectors

provision_service

Add a supported service after the demo confirmation flow

suspend_line

Suspend a synthetic line after the demo confirmation flow

get_audit_log

Inspect attempted and completed write operations

Sample subscriber IDs are sub-1001, sub-1002 and sub-1003. All phone numbers and IMSIs are reserved-looking synthetic values for this sandbox; do not use them for network operations.

Test

uv run pytest

Scope

This is a teaching and prototyping project, not a carrier emulator or production-ready agent. It models a few HLR/HSS/AuC-inspired concepts in memory and intentionally omits protocol stacks, real subscriber credentials, persistence and external connectivity.

License

MIT. See LICENSE.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

    This server cannot be deployed

    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Resources

    Related MCP Connectors

    Related MCP Servers

    • ParcelPilot MCP Server

      PraneethGoud04
      F
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to retrieve customer, order, ticket, policy, and agreement information, and to prepare or execute state-changing support actions like escalations and follow-ups with confirmation and access control.
      -

    • DungeonQ

      Ranopha
      A
      license
      Not graded
      quality
      B
      maintenance
      Enables a simulated assistant to investigate synthetic incidents, request containment actions that require separate human reauthentication and approval, apply authorized effects, and verify durable signed receipts while rejecting tampering and replay.
      Apache 2.0

    • cx-agent

      mohahesham1
      A
      license
      Not graded
      quality
      C
      maintenance
      Enables a telecom support assistant to look up customers, check balances and orders, suspend lost SIMs, open tickets, and analyze SIP traces to diagnose call failures.
      MIT