Known issue reproducer consistent

The check is default-on at warning, verify stage. It reads an active KnownIssue (open / reviewed) that states a prohibition in remediation, mitigation, tripwire_mutation, title, or description, and that carries a bound reproducer. It harvests declaration names (func / fn / def) and JS/TS it() / test() titles. A non-negated accept-verb (allow, accept, permit, promote, overwrite, clobber) that shares a subject token with the finding is the contradiction. A name that already says must_not_promote is the secure contract. It is not flagged.

This hop does not ask whether a reproducer is present and resolves. That is known issue reproducer present and resolves. It does not ask whether every active KI has some evidence kind. That is known issue complete. It does not ask whether a high-stakes KI is graded by an executed witness. That is high severity reproducer grade. A resolving selector named allows still fails here. Presence is not this floor.

id: KI-R2
status: active
title: empty api_key must be REJECTED
remediation: "must not promote the host env secret"
reproducer_command: go test ./proxy -run TestProxyAllowsEmptyApiKeyBody

The suite can still be green. The selector resolved. The test asserts the banned input is accepted. Once the defect is fixed, that test never fails. The regression net is pointed the wrong way. A KI with no bound reproducer is skipped here. Zero active KnownIssues with a bound reproducer is a pass that inspected nothing.

proof known-issue edit KI-R2 --set-reproducer-command 'go test ./proxy -run TestProxyRejectsEmptyApiKeyBody'
proof audit --check known_issue_reproducer_consistent

Rename the accept-verb so the declaration asserts the safe outcome, or declare a characterization. Do not leave an allows name on a prohibition so the hop stays quiet. A finding whose test ratifies the defect is still a finding.

The teaching graph is still one active KI next to one accept-verb declaration. Read the finding. Then rename the verb, or stamp a characterization.

// CHARACTERIZATION + MUTATION TRIPWIRE for KI-R2
func TestProxyAllowsEmptyApiKeyBody_characterization(t *testing.T) {
    // pins the current (defective) behavior
}

Close a contradiction by asserting the secure contract so the reproducer fails while the defect is open (allows to rejects, accepts to must_not_use). Close an honest pin by adding a header comment CHARACTERIZATION + MUTATION TRIPWIRE for <KI-ID>, naming the function *_characterization, and setting tripwire_mutation to the product edit that would flip it. A file that already says "assert the safe contract, fails while the defect is open" is exempt. The hop does not parse assertion bodies. A neutral-named test whose body does assert status == 200 on the banned input is not caught. Completeness stays on known issue complete. Presence stays on known issue reproducer present and resolves. High-stakes grade stays on high severity reproducer grade. Do not treat Salesforce's bulletin as this cell. Jama still authors. Proof vs Jama.

Warning severity. A counted warning does not block advancement. Rewriting the name is author judgment. The tool surfaces the contradiction. A KI with no bound reproducer is skipped, not flagged. Zero active KnownIssues with a bound reproducer is a pass that inspected nothing. Load failure is a fail. The hop is a name plus annotation heuristic, precision-first. Recall is partial: it only catches contradictions encoded in declaration names and JS/TS titles. It does not execute the reproducer. It does not inspect the implementation. It does not prove the Go. We have not scored this floor against a frozen Jama pack, a Salesforce bulletin, or a Jira export. The loss is named, not scored.

The presence hop stays on known issue reproducer present and resolves. The quality-floor hop stays on known issue complete. The high-stakes hop stays on high severity reproducer grade. The engagement stays on software correctness audit. Jama still authors.

What is known issue reproducer consistent? Same question. Same URL.

Is this known issue reproducer present and resolves? No. That hop is whether a selector, command, or Reproduces annotation actually resolves. This hop is whether that bound name contradicts the finding. See known issue reproducer present and resolves.

Is this known issue complete? No. That hop is whether every active KI has some evidence kind, including prose. This hop is whether a bound test name ratifies the banned outcome. See known issue complete.

Is this high severity reproducer grade? No. That hop is whether a high-stakes KI is graded by an executed witness. This hop is any-severity bound name. See high severity reproducer grade.

Is this a Salesforce known issue? No. Ads known issue is that status page. A hosted vendor bulletin is not this cell.

Does a missing known-issues directory fail? No. Zero active KnownIssues with a bound reproducer is a silent pass. Put a YAML there when the behavior is confirmed, then bind a test that asserts the secure contract.

Does a characterization fail? No. A header comment, a *_characterization name, or a secure-contract acknowledgement exempts the file. Pinning the current defective behavior on purpose is an honest shape.

Does a contradiction fail the merge? No. The check keeps warning severity. A counted warning does not block advancement.

Does a green hop prove the finding? No. The hop checks declaration names against prohibition prose. It does not run the command. It does not parse assertion bodies.

Is Proof a Jama alternative for the shall? No. Jama still authors. Proof vs Jama.