Status: supported for bounded repository integration.
preregistration_integrity: validremote_repository_integration: supportedpush_required_jobs: 2/2 passedpull_request_required_jobs: 2/2 passedhosted_bootstrap_observations: 4/4 passedfinal_sealed_head_checks: not_runmerged_master_verification: not_runiter241_retry: not_authorizedscientific_authority: absent
The exact candidate was
de22688f800e0fb46c15ecd851d2bf76e26b0a82, with tree
416c864123bd7451d250bcd6384c41d1670343a5. Post-reboot local validation
passed 1,589 authenticated tests with exactly two seal-qualified deselections,
and all 293 CI-declared offline commands passed. That local Darwin/Python
3.14.2 evidence did not certify hosted Linux.
Normal push run 29920504274 and pull-request run 29920506702 both
completed successfully on attempt 1. Python 3.11.15 and 3.12.13 each passed on
both events using Ubuntu 24.04 image 20260714.240.1, provisioner
20260707.563, and runner 2.336.0.
Each hosted job authenticated its exact registered archive before parsing,
retained the exact registered byte count and SHA-256, rejected group/world
write in the extracted tree, did not run the upstream setup path, and observed
the downloaded interpreter only after containment and native-loader checks.
The first-Python observations reported isolated, ignore-environment,
no-user-site, and safe-path controls; contained prefixes, pip, and libpython;
and the later authenticated test route accepted an absolute regular
owner-executable mode-0700 interpreter.
GitHub's pull-request synthetic merge was
29b5b29981c684032151ef8d6d78a88d5bb77389. Its ordered parents were live
master 6a9a4f66ec331011c9dfbe14b3a22259a5b585d5 followed by the exact candidate,
and its tree was byte-identical to the candidate tree.
The bounded machine-readable observation is retained in
proof/remote_ci_observation.json. It binds
the exact run, job, runner, archive, bootstrap, trust, and synthetic-merge
identities without retaining unrestricted logs, environment values, temporary
paths, or credentials.
The Iter245 hypothesis prospectively fixed the required evidence categories and decision rule, but it did not freeze an exact hosted-result JSON schema or semantic validator before the first push. The additive validator and known-bad fixtures introduced with this result can check internal consistency and the frozen gate identities; they are retrospective same-operator engineering checks, not independent attestation or proof of command execution by themselves. The GitHub run and job records remain the external observation source.
GitGuardian remained red for the already retained occurrences of Git commit
6a9a4f66ec331011c9dfbe14b3a22259a5b585d5 and its Git tree
76c6791ec2a051804a50f65b5297b709dea4f49c. Exact Git-object classification
supports an occurrence-specific false-positive disposition, not a general
security approval. The external check is not one of the two strict required
pull-request checks and was not waived, rerun, or hidden.
This result supports repository integration only for the exact candidate, registered archives, observed runner image, and bounded bootstrap route. It does not establish complete platform or transitive runtime provenance, independent review, independent semantic ground truth, general security, scientific truth, prevalence, detector efficacy, transfer, publication readiness, or release readiness.
Iter241 remains failed. Iter242 remains supported local engineering evidence. Iter243 and Iter244 remain failed engineering evidence. No provider or model call, scientific execution, human contact, spending, paper submission, publication, release, deployment, visibility change, or scientific authority is granted.
The result must now be indexed, synchronized across the mutable current surfaces, and sealed. Only the resulting exact sealed head may undergo fresh normal push and pull-request checks before any ready or merge decision.