Tool paths¶
A tool has to know where to find its example files, where to put the files it generates while it runs, and where to leave results the user will want later. This section answers those questions.
Note
The directories and environment variables described here are set up by the tool platform, which is separate software and is not in this repository, so they could not be checked against code. The one part that is CMS-side — the whitelist that decides which directories a launch URL may name — is covered in Directory parameter whitelist and was checked.
- Environment variables — what a session tells the tool about itself.
- Passing path variables with the invoke script — how the tool learns where its own repository is.
- Example files — where to keep the inputs you ship.
- Tool generated files — where temporary files and results belong.
For datasets too large to live near the tool, see Large data paths.
Reviewed against 2.4-main @ e097e0236d on 2026-09-10.