agentic-sdlc

Intent · CHG-2026-014920

Stage artifact from intent/CHG-2026-014920-infographic.md, header and all.

View intent/CHG-2026-014920-infographic.md on GitHub ↗
change_id: CHG-2026-014920
risk_class: R1
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 front page currently opens with a sixteen-second terminal recording. That is excellent for an engineer and it presumes the visitor already knows what they are looking at. Someone arriving cold gets gates refusing before they know what a gate is or why one should refuse.

A hand-drawn overview of the whole argument exists — the governance shift, the Substitution Test, the J-curve, the artifact chain, the autonomy matrix — on one image. It is currently a 5.8 MB PNG in the repository root, tracked by nothing.

This is the same audience gap v1.1 and the audio overview both addressed, in a third register: the reader who will neither read 898 lines nor listen for twenty-six minutes, but will look at a picture.

Who is affected

Success criteria

  1. It is the first thing on the front page, above the recording.
  2. It is legible on a phone and zoomable to full resolution.
  3. Its provenance is stated: AI-generated, from the playbook, not authored analysis.
  4. Where it disagrees with the playbook, the page says so.
  5. The repository does not carry an order of magnitude more bytes than the page needs.

The disagreement, which is the interesting part

The graphic is titled “Mapping the three planes of architecture” and lists Model Access, Context and Control. §5 of the playbook has five planes. Agent Runtime and Evidence are absent — and Evidence is a strange one to lose from a graphic whose own caption reads “evidence is a by-product”.

This repository’s entire argument is that published material must not drift from its source. Shipping “three planes” beside a playbook that says five, without comment, would be a small instance of exactly the failure the document is about. The options were to not publish it, to edit it, or to publish it with the gap named. The third is the honest one: the graphic is good, the omission is real, and a reader who notices the discrepancy should find that we noticed first.

Out of scope

Constraints