Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 12 additions & 4 deletions AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,10 @@ wrong or the expected list needs updating — decide which, do not just update t
Modules are listed in `settings.gradle.kts`; only `:testng` is published, as a shaded jar merging
the others. `docs/BUILD_SYSTEM.md` describes the structure.

Before writing documentation, read what is already in `docs/` and `.github/`. Most build topics are
covered there; a second copy drifts from the first, and this file was itself written as a fifth copy
before being cut back to links.

Two things that are not documented elsewhere and cost tool calls to discover:

- Per-project build files are named `<module>/<module>-build.gradle.kts`, not `build.gradle.kts`.
Expand Down Expand Up @@ -101,10 +105,14 @@ version heading.

## Working style

- **Verify claims before acting on them.** A statement inherited from a pull request description, an
issue or a comment is not evidence. Checking a release date or a changelog costs a minute; acting
on a wrong premise costs hours. This has already caused one full migration to be written and
reverted.
- **Verify claims before acting on them** — including your own. A pull request description, an issue
comment, a memory or a conclusion from an earlier session are all hearsay until re-checked against
whatever settles them: the build for anything about this repository, the registry or upstream
release notes for anything about a dependency. A release date costs a minute to check; acting on a
wrong premise costs hours. This has already caused one full migration to be written and reverted,
and one obsolete memory to be copied into a committed file.
- Running a documented command proves it executes, not that it does what the text claims. If the
text promises an effect — filtering, failing, producing a value — measure that effect.
- Report what you actually observed. If a run was flaky, say so and show both runs.
- When an upgrade is refused, record the error that refused it, so the next person does not retry it
blind.
Loading