get-filterable-attributes inspects an entity type and reports what can actually be filtered on, which makes the list calls usable without a schema in front of you. Session state is explicit rather than ambient: the base URL, the token and the organization each have their own call, so moving between environments is three calls instead of a restart. Credential login is marked as the last resort, with token refresh as the intended path.
A client for the PI API: session and organization setup, then category and chart operations.
- A base URL set at runtime and a connection check that reports whether the URL and authentication are both valid.
- Session handling: a token verified and refreshed, credential login as the fallback path, logout, and an organization ID applied to later requests.
- Categories listed with filtering and paging, fetched, created, updated and deleted, and their objects listed.
- Charts listed with filtering, fetched, deleted, and exported in a format you choose.
- The attributes available for filtering, discovered by examining a sample entity.
A PI API instance to point at and an account on it. Authentication is by token, with a username and password login available as the fallback.
One command plus a key — npm install -g @mingzilla/pi-api-mcp-server, then supply credentials