FounderCLI editorial operations and disclosure record

Record date: 2026-09-14. This is a first-party operational record, not independent validation. It records historical operations and the limitations found afterward. The snapshot revision identifies the code and documentation inspected; it does not imply that the current recovery has completed.

Runs 7 and 8 reached final human review and are recorded as completed. Run 7 used foundercli-editorial-v1.5 with governor-accepted releases through v1.10; Run 8 used v1.10. They produced Posts 9 and 10 from Briefs 3 and 1 and Evidence Bundles 18 and 19. Human final review approved exact Post hashes 38dc2cabfb21b13808b3b291015ebba9886b78a24dfab85694a824d46be4728b and dba3c9bcff8c5e9bd043c6ec89f82243256212b39b4465544e65a49a04aecd80. Publication decisions are historical actions, not approval of the September repair candidates.

Run 7 saved ten operations totaling 376,566 input tokens, 48,946 output tokens, and 1,227,839 active model milliseconds. A completed response whose subsequent deterministic assembly failed was absent from that older Payload ledger: 13,586 input tokens, 1,680 output tokens, and 64,307 active model milliseconds. The conservative emitted total was therefore 390,152 input tokens, 50,626 output tokens, eleven calls, and 1,292,146 active model milliseconds. Run 8 saved nine operations totaling 158,891 input tokens, 37,415 output tokens, and 858,213 active model milliseconds. These figures are model usage, not human labor or money saved.

Run 7 exposed source-ID mismatches, passage alias problems, a draft-slug ownership collision, and the omitted usage receipt. Existing completed outputs were recovered without another provider call for the source and passage fixes. Later implementations checkpoint usage before deterministic artifact handling and deduplicate replay by operation identity. The historical failures remain part of the record; recovery is not evidence that those failures never occurred.

Run 7 machine review left three advisory sourcing findings. Human final review dismissed those findings as explicitly editorial recommendations and approved the exact Post revision; the machine findings and human dispositions remained separate. Such publication dispositions are not calibration labels. The August 24 integrity audit found that Evaluation 1 lacked exact-final-candidate machine review and Evaluations 2 and 3 contained reviews only for earlier hashes. Their 57 seeded claims remain draft, audit-only records and are excluded from machine-versus-human agreement and calibration. There is no complete frozen human-label or missed-finding corpus establishing reliability for these cases.

The later quality review identified internal claim labels and insufficient article prose in the two follow-on articles. As of the September 14 snapshot, those articles are undergoing repair and the shadow-mode Playbook has not completed its current production review. Successful historical publication does not establish current quality, broad workflow generality, or readiness for autonomous publication. New pipeline, model, or policy configurations form separate cohorts; historical evaluation records do not validate v1.14.

The approved shadow-mode Brief dated August 24, 2026 records the following first-party conflict declarations. FounderCLI is itself operating a shadow-mode editorial system and has an interest in presenting that governance design as useful. The first-party FounderCLI case was designed and documented by the same publication producing this playbook. FounderCLI uses Codex and CometAPI in the system being evaluated; neither provider sponsors this article. These are the publisher declarations frozen in that Brief, not statements supplied by NIST or independent proof of agent performance.

Record basis: docs/EDITORIAL_GENERALITY.md, docs/EDITORIAL_OPERATIONS.md, docs/EDITORIAL_RECOVERY_V114_20260914.md, and the approved shadow-mode Brief with revision hash a089deb16202eadf7ecfe72f0a1a4fcb6ad8792faa65e5f97245d2581105f118. Historical Run 7 and Run 8 usage and completed states were read again from the production server on September 14, 2026. No new human evaluation or article approval was recorded to create this evidence note.