General AI policy update

[Malkovsky]

Few general AI guidance

• Less verbose code
• Less fanaticism about backward compatibility for experimental API

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-13T11:44:01.232406Z	7b16f85	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.23%. Comparing base (4e52a07) to head (7b16f85).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #68      +/-   ##
==========================================
- Coverage   91.34%   91.23%   -0.12%     
==========================================
  Files          53       53              
  Lines       13118    12920     -198     
  Branches     1830     1826       -4     
==========================================
- Hits        11983    11787     -196     
+ Misses        523      521       -2     
  Partials      612      612              

Flag	Coverage Δ
gcov	91.23% <ø> (-0.12%)	⬇️

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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Malkovsky]

@codex the PR only changes md file, why does it affect coverage?

[chatgpt-codex-connector]

Why Codecov reports a change

The Markdown-only PR does not change the executable code or its test coverage. The diff between the supplied base and head commits confirms that the only changed file is AGENTS.md—22 inserted lines.

What happened is:

1. The coverage workflow runs for every pull request targeting main, regardless of which files changed. There is no path filter excluding Markdown-only changes. .github/workflows/coverage.ymlL3-L7
2. The workflow performs a fresh coverage build and test run, then uploads the generated report to Codecov. .github/workflows/coverage.ymlL19-L28
3. Codecov compares that new head report against the previously stored report for the base commit. Those are two independently generated coverage snapshots, not a calculation based solely on the PR diff.
4. The report itself indicates snapshot variance:
   • coverable lines fell from 13,118 to 12,920;
   • covered lines fell by nearly the same amount, from 11,983 to 11,787;
   • misses actually improved from 523 to 521;
   • all modified coverable lines are reported as covered.

That pattern is not evidence that the Markdown change made code untested. It means the two uploaded gcov snapshots contained a slightly different set of instrumented/generated lines. Pixie builds coverage using the current GitHub runner, rebuilds all tests, gathers every .gcda file, and invokes gcov -pb on each one. scripts/coverage_report.shL8-L27 Header templates, compiler-generated branches, runner/compiler differences, dependency builds, or a differing set of emitted .gcda files can therefore alter the aggregate denominator between otherwise code-identical runs.

The displayed -0.12% should consequently be treated as coverage-report noise, especially because:

• no coverable source file changed;
• Codecov says all modified coverable lines are covered;
• the absolute number of misses decreased;
• the denominator changed by 198 lines.

A rerun of the base and head coverage jobs under identical conditions would likely eliminate or clarify the discrepancy. If this becomes recurring noise, the durable fix would be to make report inputs more deterministic—for example, explicitly limiting gcov collection to Pixie source files or configuring Codecov paths—rather than changing this documentation PR.

Checks performed

• ✅ git diff --name-status 4e52a070b1167ce5459ffa49da18c790708eef92..7b16f85d77caab97d0f2106605bbb7dacb09e4db
• ✅ git diff --stat 4e52a070b1167ce5459ffa49da18c790708eef92..7b16f85d77caab97d0f2106605bbb7dacb09e4db
• ✅ rg -n "coverage|codecov|gcov" .github CMakeLists.txt CMakePresets.json scripts
• ✅ git status --short --branch

No repository change was necessary, so I did not create a commit or a follow-up pull request.

View task →