[stuinfla]

What

Makes the owner-approved "story" graphics the default for the architecture and flow slots (ADR-0014), instead of the dependency-graph layout and the five-box vertical list.

• visuals.systemMap → architecture slot: inputs → engine → surfaces (with real commands) + a WORKS WITH strip + RUNS ON.
• visuals.decision → flow slot: candidates as gauges against a threshold, one skipped, one picked, animated handoff.

The rule that makes it safe

The brain does not get to assert any of it. make-diagrams verifies every claim against the cloned repo:

Claim	Check
works with: yes / partial	the verbatim quote must be in the named file
works with: no	the tool greps the repo for the terms; kept only on zero matches
decision rule	the rule.quote must be in rule.file

Unverifiable claims are dropped; if nothing survives, the slot keeps today's graph diagram (ADR-0011 — never fail a build for this).

Evidence

• 23 new unit tests (tests/story-diagrams.test.mjs), including guards proven by mutation.
• Full suite: 234 tests, 0 failing.
• Real brain run on realiti4/claude-swap: the brain authored both models unprompted (including a Codex "no"); the tool confirmed the quotes and a zero-hit search; a fabricated claim injected in a separate run was dropped (quote not found in README.md).
• Reference graphics (owner-signed-off) are expressed as data in tests/fixtures/story-claude-swap.mjs.

Known gaps (stated in the ADR)

• Mobile legibility: ~1100px canvases; relies on the page's tap-to-zoom lightbox. A portrait layout is the follow-up.
• Not measured on a corpus — two repos so far. The fallback keeps this safe; the gain is unproven until measured.
• First tools/ → src/ import; the INV-23 mutation test now mirrors the real layout.

[netlify]

✅ Deploy Preview for explainmyrepo canceled.

Name	Link
🔨 Latest commit	87ceb28
🔍 Latest deploy log	https://app.netlify.com/projects/explainmyrepo/deploys/6acaa4fec1a92c000898275c