Skip to content

Require an exact-commit CI and facts proof before publishing a release - #189

Merged
luisleo526 merged 2 commits into
mainfrom
ci/g06-exact-commit-proof
Nov 16, 2026
Merged

luisleo526 merged 2 commits into
mainfrom
ci/g06-exact-commit-proof

Conversation

@luisleo526

Copy link
Copy Markdown
Contributor

Publication now requires an exact-commit hub CI proof and a facts proof for the commit being released, before any token is minted or the release is finalized. Sweeps and facts pushes keep working behind that main-proof barrier, and the workflows that need it read repository contents explicitly.

Red first: on main, the release-workflow tests fail on the missing barrier (proof must precede the token mint; the missing proof step; the finalize step still calling the old pipeline script; missing contents: read). With this change, 267 tests across the five release modules pass with no skips, and the release dry run passes.

🤖 Generated with Claude Code

Guard tag creation, image readiness, finalization and fanout with complete
main-push CI and facts validation bound to the exact release/facts commits.
Keep the existing release identity and authorization checks, and exercise
missing, incomplete, stale and mismatched proof through the real reader.
The exact-commit main proof made every finalize sweep re-prove every
release that has a candidate, so a release that predates today's CI job
list could never pass and failed the whole sweep, dropping all consumer
deliveries. A candidate whose proof is refused but that is already
ready now takes the existing historical path (no alias, publish or
delivery for it); a refused candidate that is not ready still fails.

Also: the report-only facts freshness read is no longer proved (the
manifest write already proves the bound facts commit before any alias,
publish or ready step); a pending proof is reported as pending instead
of failed; facts-source waits up to 30 minutes on a push for its own
commit's CI, like the commit check; and the retag hint no longer
promises a retag that the barrier refuses, with the manual recovery
for a red release commit documented.

Tests: new sweep, freshness, pending and facts-source wait tests in
tests/test_release_main_proof.py fail before this change and pass
with it.

@graphify-labs graphify-labs Bot left a comment •

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Graphify reviewed this change.

Worth a look — the grounded gate found no coupling regressions or blocking issues, but 3 advisory finding(s) below merit a look before merge.

Not checked: tests were not run; formal verification proved 0 of 2 changed function(s) (2 not verified).

Formal verification. PR-changed functions: 0/2 verified (0 proven, 0 may-equivalent, 0 distinguished) · 2 not verified (1 vacuous, 1 unsupported).

Not verified on this run: \_decide\_triple (unsupported), decide (vacuous: never exercised).


Graphify review — findings

Adds a main-push proof gate, backed by a new scripts/release_main_proof.py, in front of every step that tags, publishes, finalizes, redelivers, or fans out facts, and the step fails closed when no proof exists. handle-upstream now pushes the final release commit first and waits up to 30 minutes for that commit's own proof before creating the tag, while the facts fanout waits only on push events and otherwise checks once, skipping token minting and dispatch on failure. Finalize and reconcile now run only on refs/heads/main, and the guarded proof reader and its freshness at redelivery time decide whether a stale or partially validated proof can still authorize a release.

Worth a look

  • Module ends with an incomplete expression — scripts/release_main_proof.py:318 · Escalate · high
    • unverified hypothesis — survived the adversarial refute pass; needs human review
  • Proof can become stale before redelivery — .github/workflows/release-redeliver.yml:93 · Escalate · medium
    • unverified hypothesis — survived the adversarial refute pass; needs human review
  • Concurrent asset reads can bypass candidate proof validation — scripts/release_main_proof.py:205 · Escalate · medium
    • unverified hypothesis — survived the adversarial refute pass; needs human review

Review partial — this diff was larger than one review pass covers, so later files were not reviewed; some findings may be missing.

Analysis details — impact, health, verification

Impact & health

Graphify review

Impact — 493 functions depend on the 487 functions this change touches.

Health — this change adds coupling hotspots:

  • new: main() — 1 callers, 8 callees

Verification — 493 functions in the blast radius were not formally verified this run (proofs are advisory here).

Gate & verification

graphify gate

PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.

Advisory (not blocking):

  • verification_scope: 493 function(s) in the blast radius were not formally verified this run

Test selection

Test selection

35 of 35 test file(s) selected (100%) via static blast radius.

Escalated to a full run for safety (noncode-change, uncovered-code) — the selection is not trustworthy on its own (see below). CI should run the whole suite.

  • tests/test_facts_cards.py — full-run-safety
  • tests/test_facts_descriptions.py — full-run-safety
  • tests/test_facts_fanout.py — impact, test-dependency, full-run-safety
  • tests/test_facts_validate.py — full-run-safety
  • tests/test_fingerprint_canonical.py — full-run-safety
  • tests/test_parity_bundle.py — full-run-safety
  • tests/test_parity_source_proof.py — full-run-safety
  • tests/test_release_action_pins.py — full-run-safety
  • tests/test_release_asset_fetch.py — full-run-safety
  • tests/test_release_binding.py — full-run-safety
  • tests/test_release_binding_workflows.py — impact, changed-test, test-dependency, full-run-safety
  • tests/test_release_coordinator_receipts.py — full-run-safety
  • tests/test_release_docs_runtime.py — full-run-safety
  • tests/test_release_fanout_check.py — full-run-safety
  • tests/test_release_guard_policy.py — full-run-safety
  • tests/test_release_hosted_postmerge_receipts.py — full-run-safety
  • tests/test_release_hpo_image.py — full-run-safety
  • tests/test_release_main_proof.py — impact, changed-test, full-run-safety
  • tests/test_release_main_push_inventory.py — full-run-safety
  • tests/test_release_pair.py — impact, full-run-safety
  • tests/test_release_parity.py — full-run-safety
  • tests/test_release_parity_image.py — full-run-safety
  • tests/test_release_parity_runtime.py — full-run-safety
  • tests/test_release_parity_wire.py — full-run-safety
  • tests/test_release_receipt_collect_sweep.py — full-run-safety
  • tests/test_release_receipt_legacy_previous.py — full-run-safety
  • tests/test_release_receipt_shards.py — full-run-safety
  • tests/test_release_receipts.py — full-run-safety
  • tests/test_release_redeliver.py — impact, full-run-safety
  • tests/test_release_runtime_smoke.py — full-run-safety
  • tests/test_release_sink_image.py — full-run-safety
  • tests/test_release_sources.py — full-run-safety
  • tests/test_release_workflows.py — impact, changed-test, test-dependency, full-run-safety
  • tests/test_repo_descriptions.py — full-run-safety
  • tools/release-dry-run/test_wfrun.py — full-run-safety

non-code file(s) changed (.github/workflows/facts-fanout.yml, .github/workflows/handle-upstream.yml, .github/workflows/publish.yml, .github/workflows/release-finalize.yml, .github/workflows/release-reconcile.yml …) → running the full suite for safety (a code graph can't see config/fixture/data deps)

changed code file(s) with no mapped test (.github/workflows/facts-fanout.yml, .github/workflows/handle-upstream.yml, .github/workflows/release-finalize.yml, .github/workflows/release-reconcile.yml, docs/release-main-proof.md) — a coverage gap or a missing link — running the full suite rather than only the selected tests

Selection is safe under the controlled-regression assumption; always-run tests + a periodic full run are the backstops. Advisory — it never changes the check verdict.

Docs that may be stale (advisory)

…and 10 more.

Formal verification

Could not verify: Could not verify \_decide\_triple.

The verifier did not have enough to check \_decide\_triple, so it is saying so rather than guessing. No false assurance is the whole point.

Guarantee: No guarantee either way, this is an honest abstention, not a pass.

Note: Reason: parameter `v` is annotated `Version` — outside the synthesizable primitive/collection set

Could not verify: Could not verify decide.

The verifier did not have enough to check decide, so it is saying so rather than guessing. No false assurance is the whole point.

Guarantee: No guarantee either way, this is an honest abstention, not a pass.

Note: Reason: not verifiable: all 110 sampled inputs raised on both versions — the function never executed, so 'no divergence' would be vacuous (mostly ValueError — names the real obstacle, not a sampling gap)

· 1 grounded finding(s) anchored inline below.

time.sleep(min(15, remaining))


def main(argv=None):

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Health regression — main()

fans out to 8 callees (efferent coupling).

Grounded coupling-delta finding (deterministic), not an LLM guess.

@luisleo526
luisleo526 merged commit 216e7be into main Nov 16, 2026
11 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant