docs(ci): drop the hard-coded Hydra gate count from the workflow comment

[rubenvdlinde]

What

One comment line in .github/workflows/code-quality.yml. Nothing else.

Before

      # Run the 61 Hydra mechanical gates against this PR's diff.

After

      # Run the Hydra mechanical quality gates against this PR's diff.
      #
      # Deliberately no count here: the gate set is the composer package
      # `conduction/hydra-gates` and that package is the single source of truth
      # for which gates exist and how many actually report. A digit written into
      # prose goes stale silently and then misrepresents the coverage this repo
      # has. See https://github.com/ConductionNL/.github/tree/main/hydra-gates

Why removed rather than corrected

The runner currently defines more gates than the 61 this comment claims, so the obvious fix is to bump the digit. That fix is wrong, and it is worth being explicit about why.

The count has been wrong in both directions. The prose said 61 while only 59 gates actually reported — gate-33 (axe-core) and gate-24 skip silently, so their absence looks identical to their success. Nobody noticed, because nothing in CI ever compares the number in the comment to the number in the runner. There is no mechanism by which a hard-coded digit in a YAML comment can be kept true; it can only drift, and while it drifts it overstates the coverage this repo has. Writing a fresh number in would restore the comment to accurate-for-now and re-arm exactly the same trap for the next person who adds a gate.

So the comment now names the thing that is authoritative instead of restating a fact about it. conduction/hydra-gates in ConductionNL/.github (at hydra-gates/, repo root) is where the gate list lives; a reader who needs to know how many gates run should read the package, or better, read the run output — which is the only place that distinguishes "63 defined" from "61 executed" from "59 reported".

Scope / risk

• Comment-only. enable-hydra-gates: true and the hydra-gates-ref: v1.0.1 pin are untouched, as is the surrounding rationale about check:strict being local-only and about pinning to a tag.
• code-quality.yml still parses (checked with yaml.safe_load).
• The literal 61 appears nowhere else in the repo in a gate-count sense. Deliberately left alone: #61-style issue/PR references, hydra-build.yml:60-61 line-range citations inside .claude/skills/persistence-audit/ eval fixtures, and assorted hashes/hex colours that merely contain the digits.

[github-actions]

Quality Report — ConductionNL/openbuild @ f886951

Check	PHP	Vue	Security	License	Tests
lint	✅
phpcs	✅
phpmd	✅
psalm	✅
phpstan	✅
phpmetrics	✅
eslint		✅
stylelint		✅
build		✅
composer			✅	✅ 101/101
npm			✅	✅ 654/654
PHPUnit					✅
Newman					⏭️
Playwright					⏭️

---

Quality workflow — 2026-08-04 05:26 UTC

Download the full PDF report from the workflow artifacts.

[rubenvdlinde]

Live evidence from this PR's own Hydra Gates job

Worth recording, because it turns out to be a stronger argument for removing the number than the one I opened the PR with. From quality / Hydra Gates on this very commit:

[hydra-gates] ALL 61 GATES GREEN
[hydra-gates] COVERAGE: 58 of 61 declared gates reported a result.
[hydra-gates] GATES THAT DID NOT RUN: 4 24 33
[hydra-gates] These gates skipped because a prerequisite was absent in this repo
[hydra-gates] (no manifest, no node_modules, no e2e suite, ...). They checked NOTHING.
[hydra-gates] RESULT: ALL GATES PASSED — EXCEPT GATES 4 24 33, WHICH DID NOT RUN.
[hydra-gates] This green covers 58 gates. It says NOTHING about gates 4 24 33.
[hydra-gates] Do not read it as 'the repo passes all 61 gates'.

So there is no single true number to write down:

• 61 — what the runner declares at the ref this repo pins (hydra-gates-ref: v1.0.1).
• 63 — what the package declares on main today. The comment would have been wrong the moment anyone moved the pin, and wrong in the opposite direction until they did.
• 58 — what actually executed here. Gates 4, 24 and 33 skipped for want of a prerequisite in this repo and checked nothing at all.

Three defensible numbers for one comment line, and the one a reader most needs (58) is the one that changes per-PR and can never be known statically. That is precisely why the corrected digit is not worth having.

Note also that the runner already prints all of this itself, unprompted, including the warning not to read the green as "passes all 61". The package is not just the source of truth for the count — it is the only thing positioned to tell you which gates were live on a given run. Pointing at it beats restating a snapshot of it.