[WIP] Add AGENTS.md and CLAUDE.md agent-instruction files

[sloede]

Shared agent-instruction files for AI coding tools.

• AGENTS.md — canonical, cross-tool source of truth (Codex, Cursor, Gemini CLI, Aider, …). Summarizes build/test/format, layout, architecture, conventions, and contribution norms from docs/src/.
• CLAUDE.md — thin bridge for Claude Code via @AGENTS.md import (so /code-review inherits the same conventions).

🤖 Generated with Claude Code

[github-actions]

Review checklist

This checklist is meant to assist creators of PRs (to let them know what reviewers will typically look for) and reviewers (to guide them in a structured review process). Items do not need to be checked explicitly for a PR to be eligible for merging.

Purpose and scope

• The PR has a single goal that is clear from the PR title and/or description.
• All code changes represent a single set of modifications that logically belong together.
• No more than 500 lines of code are changed or there is no obvious way to split the PR into multiple PRs.

Code quality

• The code can be understood easily.
• Newly introduced names for variables etc. are self-descriptive and consistent with existing naming conventions.
• There are no redundancies that can be removed by simple modularization/refactoring.
• There are no leftover debug statements or commented code sections.
• The code adheres to our conventions and style guide, and to the Julia guidelines.

Documentation

• New functions and types are documented with a docstring or top-level comment.
• Relevant publications are referenced in docstrings (see example for formatting).
• Inline comments are used to document longer or unusual code sections.
• Comments describe intent ("why?") and not just functionality ("what?").
• If the PR introduces a significant change or new feature, it is documented in NEWS.md with its PR number.

Testing

• The PR passes all tests.
• New or modified lines of code are covered by tests.
• New or modified tests run in less then 10 seconds.

Performance

• There are no type instabilities or memory allocations in performance-critical parts.
• If the PR intent is to improve performance, before/after time measurements are posted in the PR.

Verification

• The correctness of the code was verified using appropriate tests.
• If new equations/methods are added, a convergence test has been run and the results
  are posted in the PR.

Created with ❤️ by the Trixi.jl community.

[JoshuaLampert]

Good idea. Since I also just wrote a CLAUDE.md today for another project, I have a suggestion regarding how the agent should run julia code or tests.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.88%. Comparing base (b6b173e) to head (64dbaf6).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3087   +/-   ##
=======================================
  Coverage   96.88%   96.88%           
=======================================
  Files         647      647           
  Lines       50026    50026           
=======================================
  Hits        48466    48466           
  Misses       1560     1560           

Flag	Coverage Δ
unittests	96.88% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[JoshuaLampert]

I made a few suggestions regarding how agents could effectively run julia code.

[sloede]

@JoshuaLampert @ranocha If the PR looks good from your side, how shall we proceed? I tested this locally once and it gave me some good first results on initial tests, but I feel like there's still some fine-tuning necessary. On the other hand, if it only lives in this PR, nobody is going to test drive it under real-world conditions. Maybe just merge and invite everyone who uses such tools to test the files and continue to refine them?

[JoshuaLampert]

Thanks!

[JoshuaLampert]

We discussed that it could help to switch to TestItems.jl instead of the normal test framework of Test.jl. This would have the advantage that each @testitem (replacing @testset or @trixi_testset) can be run individually. Especially in VSCode it is nice because you can use the UI from VSCode to interact with the tests. This would also allow an agent to run only single @testitems, e.g., for computing and pasting l2 and linf errors. I use this in another repo using TestItems.jl and the workflow with the AI using a persistent julia session and explicit instructions in CLAUDE.md was pretty smooth so far. One disadvantage of the testitem workflow is that nested testsets/testitems are not supported. We would need to see if this restricts us.
Converting our current setup would also be a good job an AI agent could do. What do you think of this, @sloede?
If you want to see examples of packages using TestItems.jl, which have a similar (in the sense that most tests run some example using @test_trixi_include) test setup as Trixi.jl (although of course simpler) here are two examples: DispersiveShallowWater.jl and KernelInterpolation.jl.

[sloede]

We discussed that it could help to switch to TestItems.jl [...]

Let's move this - at least somewhat - orthogonal discussion to #3092?