[joemooney]

Metrics is a top-level desktop/mobile destination. The guide, graphic and both overview formats distinguish reported, partially covered, derivable and unsupported measures. The content covers effort touchpoints/load calibration, cycle time, traceability, drains, tokens/cost and tool reliability. Pricing limits name the two older Claude models in the pinned table; newer Claude and Codex models remain unpriced. Appendix B now describes current evidence instead of a removed table.

Rebuilt on current main (1db74dba112) after advisor feedback; only this slice's changes follow PR #3099. Pre-squash prerequisite commits were not replayed.

Validation: make check-ci-fast completed all 35 gates for this rebuilt head. Each slice's generated guide output reproduced. The final stack's guide/search/overview builds and audit reproduced in isolated author-check copies; desktop/mobile overview screenshots were inspected. No direct cargo, protected runtime changes, deployment or self-merge.

Author self-check: two fresh contexts; the first found stale human-exception wording in a linked guide, corrected in the final audit slice; the second found no remaining issues. These are author checks, not independent approval. Fresh independent review at each new head is required; the advisor integrates bottom-up.

[joemooney]

Review: PR-3100 — ✅ Approved at 25f18f9d1c4

Acceptance criterion	Status	Evidence
Desktop/mobile nav reaches Metrics; page and overview agree on cycle time (work/wait/unknown), traceability, token/cost attribution, drain outcomes, spec/review evidence, effort touchpoints, tool reliability	verified	build.py:19 adds Metrics to the shared PAGES nav (all guide pages regenerated); index.html nav and start links; styles.css mobile rule shows .nav nav at ≤760px (it was hidden before). The metrics-guide.md sections match the what-is-aida.md/.html bullet sections topic by topic
Reported, incomplete-coverage, derivable and unsupported measures are kept separate; no ROI, forecasts or live values	verified	The four headed sections in the guide and in both overview formats; the old live-number table is removed; "Not supported" names ROI, hours saved and forecasts
Numbers are fresh or dated; unknown is not zero; counts are not productivity; estimates are not human work	verified	The only numbers left are the rate-table version (2026-09-22.1, which matches data/token-rates/v1.toml), the bucket definitions and the "checked 29 October 2026" stamp. The unpriced≠zero, invocations≠productivity and ship bucket≠stopwatch caveats appear in both the guide and the overview
Command, flag and effort claims match source/help; md and html updated together; human-inclusive framing	verified	All 13 cited commands resolve with --help. --effort exists on open, pickup and pr ship ("ship/implementation touchpoint"). effort_calibration.rs:395 uses the ship bucket as actual. token_ledger.rs has no Antigravity parser. AIDA_TELEMETRY kill-switch is at usage.rs:170. Appendix items 29–31 are rewritten in both formats
Accessible original graphic and source links	verified	assets/metrics.svg has role=img, <title> and <desc>. The img alt follows the shared build.py pattern. All linked .rs, .toml and .md targets exist

CI: PR checks show only merge-hold-gate passing, because the base is the stacked -109-work branch. Locally at the head, make check-ci-fast passed all 35 fast-tier gates, and build.py regenerates the HTML with no diff.

Non-blocking (cosmetic): "prices only Claude Sonnet 4 … and Opus 4.1 …, plus their listed aliases" reads as if both models have aliases, but v1.toml lists only the claude-sonnet-4 alias. The SVG card says "Partial coverage" while the heading says "Recorded with incomplete coverage".

Merge bottom-up after the base (PR #3099) lands.