Basic FDP Usage
Getting access
FDP data on OSDF is protected by token-based authentication (SciTokens). You do not need to obtain a token by hand. fdp acquires and refreshes one for you:
This runs the consent flow and caches a bearer token per device under ~/.fdp/cache/. Tokens are short-lived, so fdp run acquires one itself when it needs to and your terminal is interactive. fdp logout deletes the cached token.
Access to DIII-D data still requires being a DIII-D user; see the note on the home page if you are not one yet.
Running your work through fdp
FDP data lives behind XRootD/Pelican, and the client libraries read their configuration when they load. The fdp wrapper sets that configuration and then launches your command in a fresh process. Run anything that touches FDP data through it:
Note the explicit python: fdp run executes the command you give it, so fdp run my_script.py only works if that file is executable and carries a shebang.
To see what it sets, or to apply it to your current shell instead:
Looking around
fdp ls /archives # list files on the origin server
fdp device list # which tokamaks are installed, and what each can do
fdp backends # which AI-assistant backends are available
Asking for data in plain language
The built-in assistant will write the pipeline for you:
fdp query "Fetch ip for shot 200000 and report the peak current in MA."
fdp chat # interactive session
See the AI Assistant page for backends, the GUI, and how to teach Claude Code, Cursor or Codex the TokSearch API.
Where to go next
- TokSearch, the pipeline framework, and its full documentation.
- Provenance and CMF, for recording what a run consumed and produced.
- Installation, including how to pin a single tokamak.