osmu-fanoutlisted

# OSMU Fanout One brief, several surfaces, one evidence trail. Fan out the content that already passed review instead of inventing a second version of the story for each channel. ## Inputs Require: - a `content-brief.json` that validates against `contracts/content-brief.schema.json`; - the draft (`draft.md`) and claim ledger (`claim-ledger.jsonl`) written from that brief; - the surfaces requested in the brief's `planned_outputs`; - locale, brand, and legal constraints for on-screen text; - the person who owns the topic and the render decision. The brief is the input contract, not something this skill writes. If it does not exist or does not validate, stop and say so rather than reconstructing the topic from the draft. ## Workflow 1. Validate the brief and confirm the draft was written from the same `brief_id`. 2. Select the claims eligible for reuse: `public: true` with a named approver. 3. Give every visual prompt exactly one claim ID as its subject, and describe only what that claim already asserts. 4. Keep the claim state in the prompt. An `inferred` or `recommended` claim may not be drawn as an observed scene, a screenshot, a chart, or a customer photo. 5. Write the video script as beats, each beat naming the claim IDs it speaks and the on-screen text it needs. 6. Carry the brief's `limitations` into both surfaces. A sample of three does not become a trend line because the format changed. 7. Record which claims were reused, which were deliberately skipped, and why,