Baseline: content coverage on the homepage (5% extract · 75% canon)

[bdelanghe]

Baseline the real content-coverage numbers for the homepage (index.html), via the vendored string-audit, as the reference point for closing the gap and tracking drift.

Measured 2026-06-23 — merged catalog (brand core + content/strings.json, 9 tokens)

Extract coverage (surface ↔ catalog): 5%

• 120 surface strings · 6 catalog-backed · 114 hardcoded · 3 unused
• The 6 matched: name + the 5 site tokens (hero-pitch, section-model, section-proof, byline, contact-cta).
• Unused on the page: tagline, description, thesis.
• ⚠ The <meta name=description> copy is hardcoded as "Bounded authority for AI agents — drawn at…" and does not match the description token — they've drifted.

Concept-drift (canon messages represented): 75%

• 3 of 4 brand-core messages land; 1 missing/drifted: the thesis "Draw the boundary at the door — a scope-bounded seam…" (best surface match 0.20).

Acceptance criteria

• Numbers recorded (here or a docs/coverage.md note).
• Agree the target coverage for the homepage — not 100%; most of the 114 are code snippets / inline <span>s that legitimately aren't copy tokens.
• Feeds the gap-closing issue + the coverage-report tooling.