RFR-2026-008: ROS profile integrity and negative-test suite

Research opportunity

Close the profile's high-severity identity, execution-packet, parsing, and referential-integrity gaps.

Background

The accepted pilot profile passes current tests, but the audit finds five high-severity controls absent.

Origin documents and trace

• research/evaluations/FE-BOUNDARY-2026-07-28/ros-fe-profile-v1-audit.md, section: PA-001 through PA-005; PA-012
• research/framework-engineering/ros-profile/open-ros-gaps.md, section: profile extension; execution manifests; mutation enforcement
• research/framework-engineering/ros-profile/README.md, section: accepted pilot baseline

Specific assumption challenged: A green current validator implies trustworthy execution and lineage records.

Supporting evidence: Provider records are not parsed, duplicate IDs pass, declared links can point nowhere, and negative tests are absent.

Unknowns

• Failure-mode coverage
• Migration compatibility
• Integrity guarantees under malformed submissions

Dependencies

• None; can begin immediately.

Suggested REP and methodology

Create REP-2026-008. Write fixture-based negative tests first; add duplicate and typed-reference checks; parse and schema-validate manifests; verify required files and declared outputs; retain human promotion authority.

Expected outputs

• v1.1 proposal
• Negative fixture suite
• Migration and compatibility report

Success criteria

Each PA-001–005 failure is caught, v1.0 fixtures remain interpretable, and no unattended canonical promotion is introduced.

Execution recommendation

• Recommended agent: research-tooling integrity agent
• Estimated effort: medium
• Expected knowledge gained: Prevents corrupt evidence lineage from masquerading as valid research.
• Frontier Score: 295
• Score inputs: knowledge gain 4/5; impact 5/5; cross-project reuse 5/5; scientific importance 3/5; dependency cost 2/5; implementation difficulty 3/5.
• Score calculation: 4 × 5 × 5 × 3 − 2 − 3 = 295.
• Score status: structured expert judgment, not measured utility.