Skip to content

diag(OMN-16788): surface per-ticket dod_verify check detail + credential capability from the sweep runner - #2938

Merged
github-actions[bot] merged 1 commit into
devfrom
jonah/omn-16788-sweep-verdict-divergence-diagnostic
Aug 27, 2026
Merged

github-actions[bot] merged 1 commit into
devfrom
jonah/omn-16788-sweep-verdict-divergence-diagnostic

Conversation

@jonahgabriel

@jonahgabriel jonahgabriel commented Aug 27, 2026 •

Copy link
Copy Markdown
Collaborator

Why

OMN-16788. The evidence-autoclose sweep records only per-ticket counts (dod_verify_total_checks / verified / failed). The per-check detail dod_verify produces goes to capture files this job never publishes, so when CI and a local run disagree on the same ticket and the same contract, the run log cannot say why.

Measured 2026-08-27, same day, same contracts:

Ticket Local CI (run 33098357673)
OMN-16682 verified 50/50, 0 failed 23/52, 29 failed
OMN-16626 verified 8/8, 0 failed 3/10, 7 failed
OMN-16690 failed 12/14, 1 failed 5/14, 8 failed
OMN-16761 verified 6/6, 0 failed 3/6, 3 failed

Which verifier is authoritative is undecided, and that question gates --apply. This is residual #2 of the 2026-08-27 beta DoD-verification status doc.

What this adds

An opt-in diagnose_tickets workflow_dispatch input. Empty is the default and the value every scheduled run uses, so the step is skipped entirely and scheduled runs cost nothing extra.

When set, it runs before the sweep (a sweep timeout cannot swallow it) and prints:

  1. What the run's GitHub credential can actually read, per repo the bound contracts cite. Capability only — no token value is printed, derived, or written. This matters because contract evidence checks are overwhelmingly gh pr view / gh pr checks / gh api repos/.../pulls/..., and _live_pr_checks_for_item is fail-closed by design: unresolvable live PR state is FAILED, not SKIPPED. A wrongly-scoped token therefore produces a confident wrong verdict rather than an error.
  2. The full per-check table for each named ticket — evidence id, status, message — plus occ_governance_ref, refresh outcome, and resolved sha.

The ticket list is passed through env and format-validated against ^OMN-[0-9]+(,OMN-[0-9]+)*$ rather than interpolated into the shell (OMN-16323).

Scope

Diagnostic only. No behaviour change to any scheduled run, no --apply, no Linear mutation, no new secret.

Note on why this lands before the run it enables

workflow_dispatch inputs are validated against the workflow definition on the default branch, so dispatching diagnose_tickets on the feature branch returns HTTP 500. The input has to exist on dev before the diagnostic run it exists to enable can be dispatched.

Refs: OMN-16788

Evidence-Ticket: OMN-16788
Evidence-Source: OCC#7322

…ial capability from the sweep runner

The evidence-autoclose sweep records only per-ticket COUNTS. When CI and a
local dod_verify disagree on the same ticket and the same contract, the run
log cannot say why, because the per-check detail goes to capture files this
job never publishes. That is residual #2 of the 2026-08-27 beta
DoD-verification status doc, and "which verifier is authoritative" gates
`--apply`.

Adds an opt-in `diagnose_tickets` workflow_dispatch input. When empty — the
value every scheduled run uses — the step is skipped entirely and costs
nothing. When set, it runs BEFORE the sweep (so a sweep timeout cannot
swallow it) and prints:

  1. what this run's GitHub credential can actually READ, per repo the bound
     contracts cite. Capability only; no token value is printed or derived.
  2. the full per-check table for each named ticket — id, status, message —
     plus the OCC governance ref, refresh outcome and resolved sha.

The ticket list is passed through env and format-validated rather than
interpolated into the shell (OMN-16323).

Refs: OMN-16788
@github-actions
github-actions Bot enabled auto-merge (squash) August 27, 2026 18:41
@onexbot-occ-writer

Copy link
Copy Markdown
Contributor

OCC autobind did not mint a companion for this PR: no changed-file candidate could be proven RED against the merge base, and emitting a PR-existence probe instead would be non-falsifiable evidence (OMN-15247). Hand-authored evidence is required.

@coderabbitai

coderabbitai Bot commented Aug 27, 2026

Copy link
Copy Markdown

Warning

Review limit reached

Next included review available in 38 minutes.

View limit details

Limit details: You’ve used the included review currently available. Your 131 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 5f28dbe3-7af6-4aa7-ad38-afd6eb70c7b6

📥 Commits

Reviewing files that changed from the base of the PR and between 27c24cd and 8fba19a.

📒 Files selected for processing (1)
  • .github/workflows/evidence-autoclose-sweep.yml

Comment @coderabbitai help to get the list of available commands.

@github-actions

Copy link
Copy Markdown
Contributor

⚠️ Hostile Reviewer — DEGRADED (informational)

Blocking findings (critical): 0
Total findings: 0
Models succeeded: none

Note: All reviewer models failed or were unavailable. Degraded results are informational during the pilot phase (OMN-8468/OMN-8524) and do not block merge. Error: all review endpoints [192.168.86.201:8000 192.168.86.201:8000 ] unreachable — preflight short-circuit (no models available)


Gate semantics (pilot phase)

Verdict Meaning Blocks merge?
passed No critical findings No
blocked CRITICAL findings found Yes
degraded All models unavailable (infra) No (pilot)

Powered by omniintelligence.review_pairing.cli_review — multi-model adversarial review (OMN-8468/OMN-8524)

jonahgabriel pushed a commit to OmniNode-ai/onex_change_control that referenced this pull request Aug 27, 2026
#7322)

* evidence: OCC companion pass 1 for OmniNode-ai/omnibase_infra#2938

* evidence: OCC companion self-bind for #7322

---------

Co-authored-by: node-occ-companion-effect <occ-companion-effect@omninode.ai>
@github-actions
github-actions Bot merged commit ef88803 into dev Aug 27, 2026
162 of 169 checks passed
@github-actions
github-actions Bot deleted the jonah/omn-16788-sweep-verdict-divergence-diagnostic branch August 27, 2026 19:25
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