Give genre its own provenance field and make data canonical

README.md

@@ -22,7 +22,7 @@ Every screen below was built end-to-end by an agent with framesmith's own vocabu
 
 ![A specialty-coffee checkout built by an agent in framesmith, light theme: a masthead with the store mark and a secure-checkout note, a form column with Contact, Delivery and Payment sections — real radio, checkbox and input controls, a two-option delivery-speed selector — beside a sticky order summary listing two line items, subtotal, free delivery, a first-order credit, the total, and a Place order button, with a gift-code card beneath. Generated from one seed (#A24B1E) on the "soft" personality and stamped with the commerce genre.](https://raw.githubusercontent.com/vicmaster/framesmith/master/docs/framesmith-checkout.png?v=2.1)
 
-> Not every screen is a dashboard. This checkout carries the `commerce` genre — the transactional counterpart to `dashboard` — so the evaluator reads its line prices and total as the design rather than as fabricated data. It still cleared the same bar as the two screens above, including designed **empty-basket** and **loading** states.
+> Not every screen is a dashboard. This checkout carries the `commerce` genre — the transactional counterpart to `data` — so the evaluator reads its line prices and total as the design rather than as fabricated data. It still cleared the same bar as the two screens above, including designed **empty-basket** and **loading** states.
 
 ```
 MCP Client → stdio → framesmith server
@@ -64,7 +64,7 @@ Run `npx -p framesmith framesmith-viewer` to start the standalone browser viewer
 
 > Above: a canvas in the detail view with the **Quality inspector** on the right — the same `canvas_evaluate` score the agent sees, with per-category bars and the issue list; the tabs along the top switch to the **Design-system** and **Feedback** panels. The toolbar exposes breakpoint previews, Compare, Fit, JSON, a **Comment** toggle for point-and-tell feedback, and lifecycle actions.
 
-**Quality panel.** The canvas detail view shows a read-only **quality inspector** on the right: the heuristic `canvas_evaluate` score (0–100), per-category bars, and the issue list — each cliché tell with its `category · tell` badge, severity, and suggestion. Issues that `canvas_autofix` can resolve carry an **auto-fixable** tag, and clicking any issue **highlights its node** in the live preview. Every gallery card also shows a color-coded score badge so weak canvases stand out at a glance. The score matches what your agent sees over MCP (same fast-mode evaluation, genre-relaxed by the canvas's preset) — it's computed for display only and never written back.
+**Quality panel.** The canvas detail view shows a read-only **quality inspector** on the right: the heuristic `canvas_evaluate` score (0–100), per-category bars, and the issue list — each cliché tell with its `category · tell` badge, severity, and suggestion. Issues that `canvas_autofix` can resolve carry an **auto-fixable** tag, and clicking any issue **highlights its node** in the live preview. Every gallery card also shows a color-coded score badge so weak canvases stand out at a glance. The score matches what your agent sees over MCP (same fast-mode evaluation, relaxed by the canvas's stored genre and applied preset) — it's computed for display only and never written back.
 
 ![framesmith point-and-tell feedback — a newsletter signup card in the detail view with Comment mode active: a popover anchored to the clicked paragraph shows breadcrumb chips (Body · Signup Card · Document · whole page) and a typed note, while the Feedback inspector tab on the right lists one open comment on "Fine Print" and one resolved comment with the agent's reply beneath it.](https://raw.githubusercontent.com/vicmaster/framesmith/master/docs/framesmith-feedback.png?v=1.8)
 
@@ -582,7 +582,7 @@ List available style guide presets. No params. Returns preset names and descript
 
 ### `apply_preset`
 
-Apply a style guide preset to a canvas. Merges preset design tokens into the canvas variables, and copies in any reusable components (`button`, `card`, `badge`) the preset defines so they can be instanced. The preset is also recorded in the canvas provenance + per-project build log. Tokens the canvas already resolves through an inherited workspace/project design system are kept rather than overwritten and reported as `preservedFromDesignSystem` — for typography this merge is field-wise (a partially-specified inherited token, e.g. just `fontSize`, is filled out with the preset's other fields rather than shadowing the whole role); set them explicitly with `set_variables` if you want the preset's values instead.
+Apply a style guide preset to a canvas. Merges preset design tokens into the canvas variables, and copies in any reusable components (`button`, `card`, `badge`) the preset defines so they can be instanced. The preset is recorded in `metadata.provenance.preset` and the per-project build log. It never changes the canvas's genre (`canvas_set_genre` owns that); a genre an older framesmith stored in the preset field is moved to `provenance.genre` first. The material preset relaxes the accent-hue and pure-black-white tells on its own, whatever the genre. Tokens the canvas already resolves through an inherited workspace/project design system are kept rather than overwritten and reported as `preservedFromDesignSystem` — for typography this merge is field-wise (a partially-specified inherited token, e.g. just `fontSize`, is filled out with the preset's other fields rather than shadowing the whole role); set them explicitly with `set_variables` if you want the preset's values instead.
 
 | Param | Type | Description |
 |-------|------|-------------|
@@ -788,14 +788,14 @@ Comments are authored in the viewer: toggle **Comment** in the detail-page toolb
 
 ### `canvas_set_genre`
 
-Durably declare what a canvas *is* — writes the evaluation genre to `metadata.provenance.preset` without the token/component churn of `apply_preset`, so an already-styled canvas can be calibrated in one call. The stamp is what `canvas_evaluate` / `canvas_autofix` / the viewer's quality panel read when no explicit `genre` param is passed.
+Durably declare what a canvas *is* — writes the evaluation genre to `metadata.provenance.genre` without the token/component churn of `apply_preset`, so an already-styled canvas can be calibrated in one call. Genre and preset are separate fields: `apply_preset` never changes the genre, and a genre an older framesmith stored in the preset field moves to its own field on the first write. The stamp is what `canvas_evaluate` / `canvas_autofix` / the viewer's quality panel read when no explicit `genre` param is passed.
 
 | Param | Type | Description |
 |-------|------|-------------|
 | `canvasId` | string | Canvas to stamp |
-| `genre` | string \| null | Genre to stamp (`"dashboard"`, `"commerce"`, `"material"`, …) — `null` clears the stamp |
+| `genre` | string \| null | Genre to stamp (`"data"`, `"commerce"`, `"material"`, …) — `null` clears the stamp |
 
-Returns `{ canvasId, genre, previous, relaxes, note?, versionHashUnchanged }` — `relaxes` lists the cliché tells the stamp relaxes; an unknown genre is stored but relaxes nothing, and the `note` says so immediately. Other provenance facts (structure, `importedFrom`) are preserved, and the stamp never moves the canvas's `versionHash` (the hash covers design content only) — recorded approvals stay valid.
+Returns `{ canvasId, genre, canonical, previous, relaxes, presetRelaxes?, note?, versionHashUnchanged }` — `canonical` is the genre with its alias folded (`"dashboard"` → `"data"`, `"checkout"` → `"commerce"`, what evaluate reports); `relaxes` lists the cliché tells the stamp relaxes; `presetRelaxes` (`{ preset, relaxed }`) appears when the canvas's preset relaxes tells on its own (the material preset), which a genre clear doesn't remove; an unknown genre is stored but relaxes nothing, and the `note` says so immediately. Other provenance facts (structure, `importedFrom`) are preserved, and the stamp never moves the canvas's `versionHash` (the hash covers design content only) — recorded approvals stay valid.
 
 ### `canvas_evaluate`
 
@@ -807,7 +807,7 @@ Open point-and-tell comments block too: when the canvas has any, the result carr
 
 A `FRAGILE` stress run blocks too. Once `canvas_stress` has recorded a verdict for the canvas, the result carries **`stress`**: `{ verdict, warnings, current }`, where `current` says whether that run was on the canvas's present `versionHash`. A `FRAGILE` verdict keeps the directive `NOT READY` even with zero evaluate issues, until a full `CLEAN` run (all five perturbations) replaces it; a subset run can record `FRAGILE` but never clear it. When `current` is false the directive tells you to re-run stress to confirm your fix. The verdict lives in server memory, so a restart forgets it. `canvas_evaluate` never runs stress itself.
 
-Whenever the `cliche` category runs, the result also carries a **`genre`** field auditing the genre decision: `{ active, source, relaxed, notRelaxed }` — the genre in effect, whether it came from the `genre` param or the canvas's provenance stamp, the tells it relaxed, and `notRelaxed` (`[{ tell, relaxedBy }]`) — tells still flagging that a *different* genre would relax. A score pinned by tells listed in `notRelaxed` usually means the wrong genre was declared.
+Whenever the `cliche` category runs, the result also carries a **`genre`** field auditing the genre decision: `{ active, source, relaxed, notRelaxed, byPreset? }` — the genre in effect (canonical: a `"dashboard"` stamp reports `"data"`, a `"checkout"` one `"commerce"`), whether it came from the `genre` param or the canvas's provenance stamp, the tells relaxed, `notRelaxed` (`[{ tell, relaxedBy }]`) — tells still flagging that a *different* genre would relax — and `byPreset` (`{ preset, relaxed }`), present when the applied preset relaxed tells on its own (the material preset relaxes accent-hue and pure-black-white whatever the genre). A score pinned by tells listed in `notRelaxed` usually means the wrong genre was declared.
 
 Whenever the `coverage` category runs on a **base** canvas, the result also carries a **`coverage`** field: `{ dataBearing, states, missing }` — whether the screen carries data-bearing content, which state variants (`canvas_add_variant`) already exist, and which demanded states are missing. It's absent on variant canvases (a variant *is* a designed state, so it doesn't demand states of itself).
 
@@ -816,7 +816,7 @@ Whenever the `coverage` category runs on a **base** canvas, the result also carr
 | `canvasId` | string | Canvas ID to evaluate |
 | `mode` | `"fast"` \| `"detailed"` \| `"llm"` | `"fast"` = JSON-tree analysis only (<100ms). `"detailed"` adds Puppeteer-based pixel-level overlap checks, laid out at the canvas's artboard size. `"llm"` runs fast-mode heuristics plus a vision-model critique (provider picked from `FRAMESMITH_LLM_PROVIDER` or whichever of `ANTHROPIC_API_KEY` / `OPENAI_API_KEY` is set — costs one paid API call per invocation). Default `"fast"`. |
 | `categories` | string[]? | Subset of `spacing`, `color`, `typography`, `structure`, `consistency`, `cliche`, `coverage`, `usability`. Defaults to all. |
-| `genre` | string? | Style that relaxes specific `cliche` gates — `"material"` allows purple and white elevated surfaces; `"dashboard"` (alias `"data"`) allows realistic figures on data-dense screens; `"commerce"` (alias `"checkout"`) allows the money on a transactional screen — cart, checkout, order confirmation, billing history. Defaults to the canvas's provenance preset if stamped (`canvas_set_genre` stamps it durably, no token churn). Pick by what the screen is *for* (read screens with published figures → `dashboard`; a screen you complete a purchase on → `commerce`; editors/admin forms → `material`; a pricing page is marketing and gets none of them), not what it contains — the result's `genre` field audits the choice. |
+| `genre` | string? | Style that relaxes specific `cliche` gates — `"material"` allows purple and white elevated surfaces; `"data"` (alias `"dashboard"`) allows realistic figures on a screen whose numbers are the user's own data (dashboards, activity, finance, fitness, logistics); `"commerce"` (alias `"checkout"`) allows the money on a transactional screen — cart, checkout, order confirmation, billing history. Defaults to the canvas's stored genre (`canvas_set_genre` stamps it durably, no token churn; `apply_preset` never changes it). Replaces only the genre: a material-preset canvas keeps the material relaxation. Pick by what the screen is *for* (read screens with the user's own figures → `data`; a screen you complete a purchase on → `commerce`; editors/admin forms → `material`; a pricing page is marketing and gets none of them), not what it contains — the result's `genre` field audits the choice. |
 
 **Categories and what they check**
 
@@ -906,7 +906,7 @@ Runs `canvas_evaluate` in fast mode and returns just the subset of issues with a
 |-------|------|-------------|
 | `canvasId` | string | Canvas to autofix |
 | `categories` | string[]? | Restrict to fixes from these categories (default: all) |
-| `genre` | string? | Style that relaxes specific `cliche` gates — `"material"` allows purple and white elevated surfaces; `"dashboard"` (alias `"data"`) allows realistic figures on data-dense screens; `"commerce"` (alias `"checkout"`) allows the money on a transactional screen — cart, checkout, order confirmation, billing history. Defaults to the canvas's provenance preset if stamped (`canvas_set_genre` stamps it durably, no token churn). |
+| `genre` | string? | Style that relaxes specific `cliche` gates — `"material"` allows purple and white elevated surfaces; `"data"` (alias `"dashboard"`) allows realistic figures on a screen whose numbers are the user's own data (dashboards, activity, finance, fitness, logistics); `"commerce"` (alias `"checkout"`) allows the money on a transactional screen — cart, checkout, order confirmation, billing history. Defaults to the canvas's stored genre (`canvas_set_genre` stamps it durably, no token churn; `apply_preset` never changes it). Replaces only the genre: a material-preset canvas keeps the material relaxation. |
 | `apply` | bool? | Write the fixes to the canvas in this call (default `false`: propose only) |
 
 **What gets auto-fixed**

VISION.md

@@ -339,7 +339,7 @@ Authoring intent: structures are scene-graph data, not prompt text — the agent
 - [x] Detectable tells (scene-graph + render): default purple / indigo accent hue, gradient / glow overuse, fake browser / phone / IDE chrome (traffic-light-dot frames), the hanging "tag-left / heading-right" header
 - [x] Honest-content check — flag fabricated-looking metrics / testimonials / logos in placeholder copy; suggest a labeled placeholder convention ("metric to confirm" + neutral block) instead
 - [x] Auto-fix ops where mechanical (swap a *known-default* purple accent → neutral, delete a fake-chrome strip), consistent with Phase 6 `canvas_autofix`; taste tells (gradient/glow, hanging header, fabricated copy) are suggestion-only
-- [x] Genre-aware loosening — `RELAXED_BY_GENRE` map; genre from `provenance.preset` or an explicit `genre` option (today `material` relaxes the purple gate)
+- [x] Genre-aware loosening — `RELAXED_BY_GENRE` map; genre from `provenance.genre` (Phase 30; originally `provenance.preset`) or an explicit `genre` option (today `material` relaxes the purple gate)
 - [x] Guidelines update — `docs/GUIDELINES.md` "Cliché & craft" section + `canvas_evaluate` / `canvas_autofix` tool descriptions steer authoring away from the tells up front
 
 _Full spec-driven breakdown in [`docs/specs/PHASE-12-SPEC.md`](docs/specs/PHASE-12-SPEC.md). Phase 12 complete._
@@ -451,7 +451,7 @@ Eight issues from one dogfooding session (a financial dashboard: #129–#136) sh
 
 - [x] Slice A — rendering primitives: per-side borders (`borderTop`…, style solid/dashed/dotted) + `strokeDasharray` on paths (#131, #132)
 - [x] Slice B — query + autofix: `find_nodes` (property/text/name → ids + paths), `canvas_autofix apply: true` + array-padding fixes (#136, #133)
-- [x] Slice C — evaluator calibration: `dashboard` genre relaxes honest-content; type-scale pinning from typography tokens; directive/llm-gate docs (#135)
+- [x] Slice C — evaluator calibration: `dashboard` genre (now `data`, with `dashboard` kept as its alias) relaxes honest-content; type-scale pinning from typography tokens; directive/llm-gate docs (#135)
 - [x] Slice D — fonts: generic aliases (mono/sans → CSS generics), stylesheet-URL family label honored, authoring-time unresolved-family warning (#134)
 - [x] Slice E — components: `create_component` (promote subtree) + `copy_nodes` (cross-canvas, defs travel) (#130)
 - [x] Slice F — charts: data-driven `chart` node (line + bar, multi-series, domains, gridlines/labels); dashboard pattern refresh (#129)
@@ -545,7 +545,7 @@ _Full spec-driven breakdown in [`docs/specs/PHASE-29-SPEC.md`](docs/specs/PHASE-
 
 The mobile attempt (2026-09-07) built one iOS screen end to end on v2.2.0 and got 100/100, READY TO PRESENT, without a single mobile check running: the touch-target rule only looks at form controls, and the screen had none. Mobile today is a screen size, not a surface. This phase gives a canvas a declared `type` (`web` or `mobile`), separates it from genre and preset so the three declarations stop colliding, and builds the three layers a surface needs: rules that judge a phone, vocabulary to build one, and starting points. Acceptance is literal: the attempt screen rebuilt from a mobile archetype, CLEAN with every tap target a primitive.
 
-- [ ] Slice A — untangle the declarations: genre gets its own field so `apply_preset` stops overwriting it, every stored canvas keeps its verdict through a legacy read, and `data` becomes the canonical purpose genre with `dashboard` as its alias
+- [x] Slice A — untangle the declarations: genre gets its own field so `apply_preset` stops overwriting it, every stored canvas keeps its verdict through a legacy read, and `data` becomes the canonical purpose genre with `dashboard` as its alias
 - [ ] Slice B — the type axis: `metadata.surface` with one device profile (`iphone-390`), `canvas_create({ type })` and `canvas_set_type`, a ruleset table keyed by type, genre and preset that composes by union and can turn rules on, a `surface` report on every evaluation, and a mobile type scale from `generate_design_system`
 - [ ] Slice C — the mobile ruleset: 44px targets as an error on every tappable primitive, safe areas, a type floor, undeclared phone-width canvases told to declare, hand-built tap targets and multi-column prose named as advisories, and stress that knows the type
 - [ ] Slice D — vocabulary: `fixed` / `sticky` positioning, `safeArea` on the root, and the `status-bar`, `home-indicator`, `nav-bar`, `tab-bar`, `button`, `list-row` and `sheet` primitives
@@ -567,7 +567,7 @@ _Full breakdown in [`docs/specs/CI-SPEC.md`](docs/specs/CI-SPEC.md). Deliberatel
 ### Issue-driven improvements (post-v1.8)
 
 - [x] `replace_matching_properties` — bulk property edit: apply one `set` to every node matching a property/value predicate (AND across keys, token refs match literally, structured values by shape) with `scope`/`type` filters and a `dryRun` preview; ends the one-`U()`-per-node grind for wide changes (issue #127)
-- [x] `canvas_set_genre` — stamp (or clear) `metadata.provenance.preset` directly, without `apply_preset`'s token/component churn, so an already-styled canvas can be calibrated for cliché-tell relaxation in one call; never moves `versionHash` (issue #162)
+- [x] `canvas_set_genre` — stamp (or clear) `metadata.provenance.genre` directly (originally the `preset` field; Phase 30 split them), without `apply_preset`'s token/component churn, so an already-styled canvas can be calibrated for cliché-tell relaxation in one call; never moves `versionHash` (issue #162)
 
 ---