[hyperpolymath]

What this is

A deliberate observation, not an adoption.

hyperpolymath/standards carries a reusable pipeline,
.github/workflows/ci-pipeline.yml, which as of today has zero callers
anywhere in the estate. Its exemption ledger,
.machine_readable/pipeline-allow.txt, is seeded empty on purpose. The
ledger's own text is explicit about how it may be filled:

An entry here must be EARNED BY OBSERVATION: the pipeline ran in that
repository and went red. It must never be seeded by inference.
[…] The honest seed is a dry-run wave: adopt the caller in a deliberate
sample, record the real verdict per repository, and add only what actually
went red.

This PR is one of five in that dry-run wave.

Expected verdict in this repository

detect counts ecosystem markers on the default branch. Measured today
across 122 tracked files:

ecosystem	markers
Rust / Nickel / ReScript / V / Haskell / Deno (supported)	0
Lean (lakefile.lean, lean-toolchain) — unsupported	2
Julia (Project.toml) — unsupported	1
Isabelle (ROOT) — unsupported	1
TOTAL (supported)	0

So the pipeline is expected to REFUSE — and to say so loudly rather than
reporting a success it did not earn. This repository has real ecosystems; the
pipeline simply does not support any of them, and detect deliberately keeps
those markers out of its denominator. The workflow states the reasoning
itself:

Folding them into TOTAL would lift it above zero and SILENCE the
zero-denominator refusal below while nothing whatsoever had been checked.
That is a vacuous gate, and a vacuous gate is worse than an honest blind
spot: the blind spot reports nothing, the vacuous gate reports success.

A red run here is the correct outcome and is exactly what this wave is for.

Known defect, disclosed up front

⚠ metadatastician/standards#976 — detect counts *.ncl into its ecosystem
denominator, and the RSR template ships a six-file k9 Nickel set into every
seeded repository. TOTAL can therefore be lifted above zero by template
boilerplate alone, which silences the zero-denominator refusal while nothing
meaningful has been checked. On a 24-repo sample, 9 would go vacuously
green that way.

This repository was chosen precisely because its verdict does not depend on
that contaminated signal. It has zero .ncl files, so TOTAL is 0 whether or not the Nickel probe is corrected. The observation therefore survives
whichever cure for #976 is chosen.

Deliberate choices

• Pinned to an immutable commit (2eb9bb19, ci-pipeline.yml blob
  d2583657), never @main — following main would let one push in
  standards change the verdict of every already-pinned caller with no review
  here.
• No secrets: block at all. ci-pipeline.yml declares
  workflow_call.inputs only, has no secrets: block and references no
  secrets.* anywhere. secrets: inherit would hand the entire store to a
  workflow that consumes none of it.
• pull_request + workflow_dispatch only, so the PR itself is the
  observation.
• actions.lock edited by hand, never gh actions-lock rewrite mode. Verified: the estate verifier returns rc=0 on the modified tree, so this repository stays lock-green.

Please do not merge yet

The verdict is the deliverable. Once the run completes, the result is recorded
against the ledger, and only repositories that actually went red are added
— in a commit carrying a column-0
Ratchet-exception: .machine_readable/pipeline-allow.txt — <why> trailer.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Ji1bq3TypfycfUPAR7hSxR