Rules and Evidence Model

Ten rules

Each rule is a policy. The table separates what code enforces, what is guidance, and what is not implemented, so no rule is overstated.

RulePolicyEnforced by codeGuidance onlyTestedNot implemented
INT-001Human approvalapprove/attest need reviewer, note, exact revision, --confirm-humanwho counts as a legitimate approveryesverified identity
INT-002Revision-bound evidencelink rejects stale revision; old links kept as history; AUD-05migration of evidence across revisionsyesautomated migration
INT-003No completion by proxyno completed state; attest needs four evidence types; AUD-04, AUD-11wording in downstream toolsyesindependent execution proof
INT-004Untrusted content boundarysources labelled untrusted; text never executed; size capsagent behavior in IDEspartlyhost-level sandboxing
INT-005Data minimizationlocal only; heuristic redactionwhat teams choose to importpartlyDLP-grade detection
INT-006Read-only defaultAPI is GET-only; CLI writes only .intentsutra/ and chosen report dirsconnector write policyyesconnectors
INT-007ProvenanceSHA-256 of sources and evidence; provenance labels; AUD-06, AUD-09reviewer disciplineyessignatures
INT-008Unknown stays unknownquestions; [TO CONFIRM]; unknown outcomesstakeholder follow-upyessemantic gap detection
INT-009Independent verificationaudit re-derives state from raw data and cross-checks the logan auditor outside the teamyesverified test execution (M1)
INT-010Truthful publishingdocs state limits; labels in reportsrelease reviewpartlyautomated claim linting

Evidence kinds

KindRecordedHashedWhat it can support
codeworkspace fileyes"this file existed and had these bytes"
testworkspace fileyes"a test file exists" (not that it passed)
test-runworkspace file + outcomeyes"a human imported this result" (not independently verified)
releaseworkspace file + outcomeyes"a human imported this release claim"
ticketreference textnoa pointer
prreference textnoa pointer

Trust labels

Future schemas separate observed-file, imported-claim, self-attested, measured-test, verified-external-signature, unknown, not-run and stale. This release implements the first three plus stale; measured-test and verified-external-signature are planned and are never shown because the code cannot substantiate them.

Stale evidence

When a revision changes canonical content (title, actor, trigger, behavior, constraints, acceptance), older approvals, links and attestations remain in the record for that older revision and no longer count. Reports show the stale count. Nothing is migrated silently.

The attestation gate

attest succeeds only when, for the current revision, there is: an approval, an observed code link with matching hash, an imported passed test-run, and an imported released release claim. The reviewer then confirms by hand. The state name human-attested-evidence-linked deliberately avoids "complete".

Independent audit checks

IDCheckSeverity
AUD-01audit hash chain intactcritical
AUD-02content hashes match stored fieldshigh
AUD-03evidence files still match recorded hasheshigh
AUD-04no false completion (attested state re-derived)critical
AUD-05approvals bound to existing revisionshigh
AUD-06imported claims keep "not independently verified" labelhigh
AUD-07approvals never claim verified identityhigh
AUD-08registry state is explained by audit eventscritical
AUD-09sources untrusted-labelled and hash-pinnedmedium
AUD-10stale evidence visible (informational)info
AUD-11proof-by-proxy gaps (test file without pass, PR without code, approval with open questions)medium