change_id: CHG-2026-014918
risk_class: R2
autonomy_tier: A2
controls: ['TRC-01', 'CHG-04']
data_classification: internal
originator: olaf@freundcloud.com
agent_identity: svc-agent-platform
model_route: gateway/tier-frontier
supersedes: None
Problem
The playbook is now 898 lines. v1.1 added a plain-terms front page, a glossary and a line of plain English under every heading precisely because half its intended audience — board members, business sponsors, risk officers without an engineering background — were bouncing off §5 and never reaching §6.
Readability was the right fix and it does not reach everyone. Some of the audience will not read a long document in any register, and will listen to a twenty-six minute conversation on a commute.
A Google NotebookLM audio overview of the playbook exists. It is currently a 50 MB file in the repository root, tracked by nothing, claimed by no plan, and refused by TRC-01 — which is how it was noticed.
Who is affected
- The reader who would engage with this material but not in this format.
- Anyone cloning the repository, who pays for however this file is stored, permanently.
Success criteria
- The audio plays from the site, without a download and without leaving the page.
- It is discoverable: named in the navigation, on the home page and in the README.
- It is honestly labelled as AI-generated narration derived from the playbook, not as authored commentary, and the playbook stays the source of truth.
- The repository does not carry four times more bytes than the content needs.
Out of scope
- A transcript. Worth having for accessibility and search, and it is a separate piece of work with its own verification — a generated transcript nobody has checked against the audio is a second source of truth that disagrees with the first.
- Playback that survives navigating between pages. This is a static multi-page site; audio stops on navigation. Making it persist means client-side routing, which is a large change to a small site for a small benefit.
Constraints
- Git LFS is not an option. GitHub Pages does not resolve LFS objects; it would serve the pointer file and playback would break. The audio has to be a real file in the repository, and therefore in its history permanently.
- Whatever ships is permanent. That makes the encoding decision worth making once, properly.