Summary

The rate limit decides how you use it.

The public API allows one request per 20 seconds, and that budget is shared across every user of the public key — so live querying in a conversation will stall. The maintainer's own advice is the right shape: pull what you need on a periodic job, store it, and refresh weekly. Also note the coverage boundary: the public tier exposes the current schedule and those published in the past 12 months, nothing older.

What it is

A wrapper around the public PBS Data API, the Australian government's record of subsidised medicines. The assistant names an endpoint and parameters; the server handles the subscription key, rate limits and error cases.

What you get

One tool taking an endpoint, an HTTP method, query parameters, an optional subscription key and a timeout — the whole public API through a single call shape
Endpoints include prescribers, item-overview, items, schedules, atc-codes, organisations, restrictions, parameters, criteria, copayments, fees, markup-bands, programs and summary-of-changes
Both stdio and HTTP with SSE, so the same build serves a desktop client and a web one
A command-line mode with subcommands for listing endpoints, fetching API information, querying prescribers and item overviews, and running an arbitrary endpoint — useful for a scheduled cache refresh
Error handling that distinguishes rate limiting from authentication failures

Requirements

Node — clone, install and build, then point the client at the compiled entry point; published as pbs-mcp-standalone, version 1.0.0. A PBS API subscription key goes in PBS_API_SUBSCRIPTION_KEY, obtained by registering at the PBS Data API Portal and subscribing to the PBS Public API v3 product.

Setup effort

Build from source — clone the repository and build it, then point your client at the binary