Public launch proof

This page demonstrates two current repository capabilities:

1. reduce a saved CI failure to the first meaningful failure, affected file, focused proof command, and review-first decision;
2. profile an external-repository fixture without installing dependencies, executing target code, mutating the target, or authorizing a merge.

Published package versus main

This proof records repository main behavior at source commit f367b25b003efebb75dcaa72fd229979be59b8c2. These capabilities remain main-only until the qualified 1.1.0 release. The published 1.0.3 wheel is not proof of these surfaces.

Reproduce the proof

python scripts/build_public_launch_proof.py \
  --source-commit f367b25b003efebb75dcaa72fd229979be59b8c2

Committed repository evidence:

docs/artifacts/public-launch-proof/failure-diagnosis.json
docs/artifacts/public-launch-proof/adoption-story.json
docs/artifacts/public-launch-proof/walkthrough.md

The contract test regenerates those files and fails when committed evidence drifts. They remain repository evidence rather than built-site pages because docs/artifacts/** is intentionally excluded from MkDocs.

Failure diagnosis

Human-readable excerpt from tests/fixtures/public_failure_demo/ci_log.txt:

Run python -m pytest -q tests/test_checkout.py
FAILED tests/test_checkout.py::test_total_includes_tax - AssertionError: calculated total was 108; expected 110
=========================== 1 failed, 48 passed in 1.42s =======================
Process completed with exit code 1

The committed JSON preserves the exact fixture line. This page uses equivalent descriptive wording so repository hygiene scanners do not mistake documentation for a weak live-code assertion.

Exact operator command:

python -m sdetkit investigate failure \
  --log tests/fixtures/public_failure_demo/ci_log.txt \
  --format json \
  --out build/public-launch-proof/investigation.json

Field	Evidence
Ecosystem	python
Tool	pytest
Classification	test
Investigation classification	PYTEST_ASSERTION_FAILURE
First meaningful failure	tests/test_checkout.py::test_total_includes_tax reported calculated total 108; expected 110
Affected file	tests/test_checkout.py
Focused proof	PYTHONPATH=src python -m pytest -q tests/test_checkout.py::test_total_includes_tax -o addopts=
Review first	true
Safe fix allowed	false
Merge authorized	false

The system diagnoses and recommends proof. It does not claim that a patch is safe, equivalent, or merge-authorized.

Fixture-based adoption story

Target fixture path: tests/fixtures/public_adoption_target/

Surface	Result
Languages	go, javascript_typescript, python
Package managers	go_modules, npm, pip
Test runners	node_test_script
CI	gitlab_ci
Security	pip_audit
Recommended manual proof	go test ./..., npm test
Review-first unknown	Python project detected but test command is not proven

read_only=true
dependencies_installed=false
target_code_executed=false
target_repository_mutated=false
automation_allowed=false
patch_application_allowed=false
merge_authorized=false
semantic_equivalence_proven=false

The recommended commands were not automatically executed against the target.

Accessibility description

This is a text-first static walkthrough. It shows a pytest failure log, the first failing node, the affected file, the focused proof command, detected mixed-stack surfaces, and explicit false authority fields. No information depends on color, animation, or an image.

GitHub About field source

Deterministic release confidence for CI: first-failure diagnosis, ship/no-ship evidence, and review-first proof commands.

1.1.0 proof-led announcement draft

SDETKit 1.1.0 turns saved CI evidence into a first meaningful failure, affected file, focused proof command, and explicit review-first decision. Its adoption profiler identifies Python, JavaScript/TypeScript, Go, CI, and security surfaces without installing target dependencies, executing target code, changing the target repository, or authorizing a merge.

This remains a draft until the qualified 1.1.0 release workflow succeeds and the public package is verified.