Governed calibration on pypa/packaging

Apex has completed governed calibration on a private, governed mirror of software it does not own - no outside custody is claimed - according to its retained calibration account. The public signed projection reports a bounded result; primary receipts are not public. Its integrity can be checked locally, but this page does not reproduce the underlying evaluation.

This is a historical account, not a live feed. New claims require source evidence and a separate disclosure decision.

Packaging - calibration lane

The target is pypa/packaging, the library behind Python's version and dependency semantics, run as a private governed mirror. Apex has completed governed calibration against the dimension evaluator:

• The earlier run floored operations.retry_bound.^{evidence summary}
• The signed summary records performance.nesting_ceiling from FAIL to PASS while naming 11 protected dimensions.^{evidence summary}What this does not proveThe public projection is a signed summary. Its signature check does not rerun the acceptance verdict.Public trailSigned summary ↗
• Target-native test result in the signed summary: the packaging test suite itself - 26,854 passed, zero failures, errors or skips.^{evidence summary}The suite is the target's own pytest suite, run unmodified. Apex did not write the test it is judged by.
• The fresh-clone check is a same-engine re-evaluation, recorded privately: it re-derived the evaluation digest and the recorded verdicts from committed inputs. It is not independent verification.^{evidence summary}What this does not proveThe fresh-clone replay is reported in a private record; a visitor cannot reproduce it from this page alone.Public trailReplay / evidence record ↗
• The signed projection records authorityState AUTHORITATIVE and verdict COMPLETED; floor ADMITTED; acceptance PASS. It does not name a person. The public projection is signed; no signature over the private primary receipts is supplied here. Its acceptance field does not establish native Apex Review ACCEPT.Public artifact integrity and the authority of a private acceptance decision are separate questions.

sourceRevision : string
baseline       : string
targetDimension: string
beforeAfter    : { before: string; after: string }
protectedDimensions: string[]
nativeTestResult: string
evidenceDigest : string
verdict        : string
authorityState : string
lane           : string
freshness      : string
timestamp      : string
signature      : string

What this is: a calibration lane. The receipts are marked development-only. The public summary reports a bounded result, not independent reproduction of the floor and ratchet machinery. These records are not a sealed fix of a real production deficiency, and they are not custody of anything.

Astronomy Shop - historical feasibility

The OpenTelemetry Astronomy Shop, a deliberately messy microservices demo, was Apex's initial external target. Early campaigns exercised the full campaign loop - observation, proposal, evaluation, gated acceptance - against a live distributed system. We classify those runs as historical feasibility evidence: genuinely informative about the machinery, not yet a reproducible public proof campaign. Whether Astronomy Shop becomes an accepted public proof, or stays demoted to history, is an open decision. We will document the outcome here.

Receipts and repositories

The underlying receipts, governed mirrors, replay tooling and full evaluation records remain private. No authorized access procedure is established here; a future disclosure would require separate authorization.