[Critsium-xy]

Reminder

• I have read AGENTS.md and docs/developers_guide/agent_governance.md.
• I have linked an issue or explained why this PR does not need one.
• I have added adequate unit tests and/or case tests, or explained why not.
• I have listed the exact verification commands run and their results.
• I have described user-visible behavior changes, including INPUT parameter changes.
• I have explained core-module impact for ESolver, HSolver, ElecState, Hamilt, Operator, Psi, or other source/ changes.
• I have requested any needed governance exception below.

Linked Issue

No linked issue. This is a documentation-only governance change, prompted by a pattern seen in recent PRs (see below).

Unit Tests and/or Case Tests for my changes

• Commands run:
  • git diff --cached --check
  • python tools/03_code_analysis/agent_governance_check.py --staged
• Result summary: no whitespace errors; governance check reported no findings. Both files have LF line endings (git ls-files --eol).
• Checks not run, with reason: no build or CTest run. The PR changes only Markdown and adds no mechanical check.

What's changed?

Integration tests in ABACUS are Autotest cases: a case directory with INPUT/STRU/KPT and result.ref, listed in CASES_*.txt and run by Autotest.sh. Several recent PRs (#8053, #8069, and the open #8235 and #8242) instead register standalone Python scripts with add_test(). These scripts launch ABACUS, build or rewrite inputs at run time, and assert on the output themselves. Each PR has invented its own launcher, environment, and fixture handling. Examples:

• One driver rewrites another suite's INPUT by exact-whitespace string replacement.
• One uses find_package(Python3 ... REQUIRED), which makes Python a hard configure-time dependency for MPI test builds.
• One drops MPIEXEC_PREFLAGS/POSTFLAGS and the PROCESSORS property.

The written rules said nothing about this, so reviewers had no basis to push back.

This PR:

• Adds AGENTS.md rule 22:
  • Integration tests are Autotest cases by default.
  • A standalone add_test() driver is allowed only for checks Autotest cannot express, such as an expected nonzero exit or per-rank stderr.
  • Python helpers called from Autotest (e.g., CompareFile.py) are unaffected.
• Adds an "Integration Test Drivers" section to agent_governance.md with the requirements for a justified driver:
  • It lives in tests/integrate/validation_tools/ and is named check_*.py.
  • Its inputs come from a dedicated fixture directory, with no run-time rewriting of another case's INPUT.
  • The interpreter is optional: find_program plus if(), never REQUIRED.
  • MPI drivers pass through the launcher flags and set PROCESSORS.
  • It sets OMP_NUM_THREADS=1, uses a temporary scratch directory, and sets a CTest TIMEOUT.
  • The PR states why Autotest is insufficient.
• Adds a matching rule-grading matrix row, set to AI review + human confirmation. No mechanical blocker is introduced, and existing drivers are not affected retroactively.

Governance Notes

• INPUT/docs changes: developer documentation only (AGENTS.md, docs/developers_guide/agent_governance.md). No INPUT or user-facing behavior change.
• Core module impact: none; no source/ changes.
• Exceptions requested: none.

🤖 Generated with Claude Code