Macro-5 Staging Object: Run Candidate — R2-B2 (2026-06-19)

Date: 2026-06-19 · Workstream: R2-B2-MACRO-5-STAGING-BUILD-AUTHORIZATION-PACKAGE-2026-06-19 (Deliverable 23 of 110) · Editorial revision: rev1 Class: staging object candidate (run) · READ-ONLY · NON-ENACTING · NON-AUTHORIZING · NO write performed.

Metadata convention. Editorial revision (rev1) only. AgentData storage revision/content_length authoritative at read time.

NON_EXECUTABLE_TD_CANDIDATE. Conceptual object only. Do not create a real object.

---

0. Status and non-authorization

STATUS: PASS — engineering / design-only. The run conceptual object: one disposable producer run. Engineering PASS ≠ authority PASS. Default: HOLD.

1. Purpose

Describe the run object as the scope unit of a workbench exercise.

2. Sources / evidence read

Object model (14); Macro-4 staging workbench lifecycle (D8); B2 evidence contract (D32, run identity). Main process, no reader-agents.

3. Accepted baseline (carried)

One producer run = one rollback/delete unit; a run carries its own id, channel id, and rule-set version.

4. Evidence / analysis — run candidate

Aspect	Description (no schema)
role	one disposable end-to-end exercise of a candidate producer
field family	run-id, channel-id, rule-set-version/hash, start/end, status
scopes	the delete-fast unit (all artifacts of this run)
never	spans production rows; reuses a production id

5. Candidate / requirement / gate / result

The run object is the anchor for disposal and evidence. It is a role, not a table; no run is created or executed. A run that cannot enumerate its own artifacts is defective (residue risk, BAD-15).

6. Owner-gated future work

Creating/executing a run is Owner-gated; forbidden now.

7. What remains unresolved

The concrete run-id format and store are FUTURE_TECHNICAL_DESIGN_REQUIRED.

8. Ready for GPT/Codex review

Yes — Codex should confirm the run object stays conceptual and scopes disposal.