@erclx/canon 4.110.0 → 4.111.0

package/claude/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "canon",
   "description": "Automated governance, versioning, and discovery tools for Claude Code.",
-  "version": "4.110.0",
+  "version": "4.111.0",
   "author": {
     "name": "Eric Le",
     "url": "https://github.com/erclx"

package/claude/skills/canon-operator/REQUIREMENT.md

@@ -26,8 +26,8 @@ The last failure is a section no route reaches. `## Route` maps an intent or a d
 - Resolve every stack, rule, snippet, and standard name from a catalog at runtime
 - Run the CLI non-interactively and report the command run, what changed, and the full path of anything written
 - Read the overwrite contract from the body before any install or sync, and name each surface it lists as overwritten before running one
-- Read the reverse walk beside the forward sections, so a folder the toolkit stopped shipping and a migration with a live case both reach the user
-- Hand off a named migration to the skill the report names, and name an unmigrated domain's root and install path for the user to move themselves when no skill covers it
+- Read the reverse walk beside the forward sections, so a folder the toolkit stopped shipping reaches the user
+- Name an unmigrated domain's root and install path for the user to move themselves, since no skill covers the move
 - Route a measurement intent to the audit offers from the route table itself, rather than leaving the section reachable only by reading past the table
 - Rank a lifecycle row against the audit offers where the route table itself states it, so a session acting on the lifecycle row reads the ranking without opening the section being ranked
 - Offer every audit whose surface the target carries, and offer none whose surface it lacks

package/claude/skills/canon-operator/SKILL.md

@@ -25,10 +25,10 @@ Run `canon sync --check . --json` before routing. It reports what a target is be
 Read seven fields off the report and carry each to `## Route`:
 
 - `unmigrated`: a domain sitting at the root layout with nothing under `.claude/`. The most urgent finding, because that domain reports no drift of its own while being entirely behind
-- `superseded`: a file a newer seed folder replaced. No command moves it, since the content is the project's own, so this routes to the skill that proposes the split rather than ending at the report
+- `superseded`: a file a newer seed folder replaced. No command moves it, since the content is the project's own, so this routes to `canon:target-check` for the full per-domain reading rather than ending at the report
 - `seeds`: entries are `matching`, `stale`, `drifted`, or `missing`. Anything but `matching` needs the seed handoff
 - `domains[].entries`: per-file `stale`, `customized`, `stranded`, and `orphaned` as before
-- `reverse`: what the target holds that no live catalog claims. `unclaimed` lists folders at roots the toolkit stopped shipping, each with an `attribution` of `dropped`, `unattributed`, or `project`. Act on the first two and leave `project` alone, which history proved the project owns. `migrations` names a proposal-only skill with a live case here and the `reason` it was measured from
+- `reverse`: what the target holds that no live catalog claims. `unclaimed` lists folders at roots the toolkit stopped shipping, each with an `attribution` of `dropped`, `unattributed`, or `project`. Act on the first two and leave `project` alone, which history proved the project owns
 - `historyUnavailable` on a domain, on `seeds`, or on `reverse`: nothing could be dated, so treat every difference as unverified and say so rather than reporting a file as untouched. It is set on `reverse` when the toolkit itself ships without history, which is the registry-install case, so the walk found nothing rather than finding a clean target
 - `tooling`: read `measured` first. Every count under it is zero when it is false, which is an absence of measurement rather than a measured zero. Past that, `chain` names the stacks the install resolved, nearest first, and `counts.gitignore` counts the managed ignore entries the target is missing.
 
@@ -43,8 +43,7 @@ Map the stated intent, or what `## Diagnose` found, to one lifecycle phase, then
 The two can name different rows, and a reply answers both rather than picking one. Run or hand off what the finding names, then answer the intent in the same reply. A measurement intent therefore reaches the audit offers below even on a target whose diagnostic found scaffold work, which is the case a session leaving from the scaffold row would otherwise never read.
 
 - A domain in `unmigrated`: name it, the root path, and the install path. No command moves it, so the user runs the relocation themselves
-- An entry in `migrations`: hand off to the skill its `skill` field names, spelled as the report spells it. Both of them propose without writing, so the handoff is where this stops
-- Anything in `superseded`: name which files and what replaced them, then hand off to `migration-superseded`. Both skills propose, so neither moves nor deletes the file
+- Anything in `superseded`: name which files and what replaced them, then hand off to `canon:target-check`, which reports and never moves or deletes the file
 - A folder in `unclaimed`: name it and the attribution it carries. No command moves it and the content may be the project's own, so the decision is the user's
 
 - First-time scaffold of a fresh project: hand off to `target-setup`

package/claude/skills/target-check/references/domains.md

@@ -25,7 +25,7 @@ Read three fields off the same report, then two verbs:
 
 - `unmigrated`: a domain still at the root layout with nothing under `.claude/`. The most urgent finding, because such a domain reports no drift of its own while being entirely behind.
 - `superseded`: a file a newer seed folder replaced. No command moves it, since the content is the project's own.
-- `reverse`: `unclaimed` names folders at roots the toolkit stopped shipping, each with an `attribution` of `dropped`, `unattributed`, or `project`. `migrations` names a proposal-only skill with a live case here and the `reason` behind it.
+- `reverse`: `unclaimed` names folders at roots the toolkit stopped shipping, each with an `attribution` of `dropped`, `unattributed`, or `project`.
 - `canon context audit --json`: required sections, entry length, citations, and index drift across the context folder.
 - `canon records validate <kind> --json`, once per record folder the target carries.
 
@@ -33,7 +33,7 @@ Read three fields off the same report, then two verbs:
 - Behind: any of those carrying an entry. Name the entry and its attribution.
 - Unread: `historyUnavailable` set on `reverse`, a `reverse` key the report does not carry at all, or either verb refusing. A missing `reverse` key is a binary predating the field rather than a target with nothing unclaimed.
 
-Routes by finding. `unmigrated` routes to a relocation the user runs, since no command moves it. `superseded` routes to `migration-superseded`. A `migrations` entry routes to the skill its `skill` field names, spelled as the report spells it. An `unclaimed` folder carrying `project` attribution routes nowhere, since history proved the project owns it.
+Routes by finding. `unmigrated` routes to a relocation the user runs, since no command moves it. `superseded` routes to a split the user makes, since no command moves the content and no skill proposes it. An `unclaimed` folder carrying `project` attribution routes nowhere, since history proved the project owns it.
 
 ## 3. Governance rules
 

package/docs/agents/install-and-sync.md

@@ -256,11 +256,9 @@ seed tree rather than from a fixed set of filenames, so a folder added later is
 covered without a code change. Only an exact stem matches, which leaves a
 suffixed variant such as `TASKS-ARCHIVE.md` unreported.
 
-Route it to `migration-superseded`. That skill resolves the standard governing
-`replacedBy` from the `appliesTo` the standards catalog declares, reads the
-destination shape with `canon standards <name>`, and proposes the split without
-writing. Where a folder has no governing standard, the entry earns a named
-refusal rather than a shape nobody stated.
+No skill splits the file. The standard governing `replacedBy` is the one the
+`appliesTo` the standards catalog declares, and `canon standards <name>` reads
+its destination shape, so the split is the project's own to make.
 
 `unmigrated` names a domain sitting at the root layout an older toolkit installed
 to, with nothing at the path the current one reads. It carries `rootPath`,
@@ -350,7 +348,7 @@ check.
 `reverse` is the one section built by walking the target rather than the
 catalog. Every other surface enumerates toolkit-owned keys and asks whether the
 target matches, so a folder the toolkit deleted appears in none of them. It
-carries `unclaimed`, `migrations`, and `historyUnavailable`.
+carries `unclaimed` and `historyUnavailable`.
 
 `unclaimed` names a folder the target holds at a top-level path the toolkit once
 shipped and has since deleted. The candidate roots come from the toolkit's own
@@ -381,13 +379,6 @@ colliding on a retired name costs the walk no reads. The cost is that a file the
 toolkit shipped and the target renamed goes unmatched, the same limit the
 `unmigrated` count carries.
 
-`migrations` names a proposal-only skill with a live case in this target. It
-fires on a `CLAUDE.md` past 250 lines for `migration-claude-md`, and on a `docs/` folder
-holding markdown with no populated `canon/context/` for `migration-context`.
-Each entry carries the skill name and the measurement behind it, so a consumer
-can check the proposal before running it. Without the field both skills are
-documented and unreachable from any report.
-
 #### What counts toward the gate
 
 `unmigrated` counts toward `--exit-code`, since running the relocation closes it.

package/docs/target-projects.md

@@ -261,8 +261,6 @@ Each entry carries a verdict, since a dropped folder and one the project wrote a
 
 Nothing acts on any of them, and the verdict is what makes the list safe to read.
 
-The same field names a proposal-only skill with a live case here under `migrations`, which is how `canon:migration-claude-md` and `canon:migration-context` become reachable. Each entry carries the measurement behind the proposal rather than the proposal alone.
-
 Add `--json` for the machine-readable report, and `--exit-code` to fail a CI job when a target falls behind. Files the project authored itself never count toward that exit code, and neither do superseded artifacts, seed drift, tooling, or anything the reverse walk reports, since each names content the project is expected to edit or place itself. An unmigrated domain does count, because running the relocation closes it.
 
 Tooling reports under a section of its own, and `measured` there says whether the target ever recorded a chain. One that never ran a tooling sync reports unmeasured rather than clean, which is what separates tooling nobody has looked at from tooling that is current. A workspace root records nothing either way, since each package resolves its own chain.

package/docs/workflow/ai-workflow.md

@@ -285,17 +285,13 @@ This section is the corpus the coverage claim is measured against: every name `c
 
 ### Keep the project current with the toolkit
 
-| Skill                            | When to use                                                                            |
-| -------------------------------- | -------------------------------------------------------------------------------------- |
-| `canon:target-check`             | In a target project, to report per domain what it holds against what the toolkit ships |
-| `canon:seed-sync`                | After a toolkit update, to reconcile installed seeds without losing customizations     |
-| `canon:migration-claude-md`      | When `CLAUDE.md` grew past what always-load context should carry                       |
-| `canon:migration-context`        | When `docs/` holds agent-flavored files belonging in `canon/context/`                  |
-| `canon:migration-superseded`     | When a drift report names a `.claude/` file a folder has replaced                      |
-| `canon:migration-standards-drop` | When the project still holds an installed `.claude/standards/` tree                    |
-| `canon:canon-feedback`           | When something in the toolkit is broken, missing, or off, to open an issue on it       |
-| `canon:canon-feedback-triage`    | In the toolkit repo, to work through the open feedback issues                          |
-| `canon:canon-rollout`            | In the toolkit repo, to take one change out to every consuming project at once         |
+| Skill                         | When to use                                                                            |
+| ----------------------------- | -------------------------------------------------------------------------------------- |
+| `canon:target-check`          | In a target project, to report per domain what it holds against what the toolkit ships |
+| `canon:seed-sync`             | After a toolkit update, to reconcile installed seeds without losing customizations     |
+| `canon:canon-feedback`        | When something in the toolkit is broken, missing, or off, to open an issue on it       |
+| `canon:canon-feedback-triage` | In the toolkit repo, to work through the open feedback issues                          |
+| `canon:canon-rollout`         | In the toolkit repo, to take one change out to every consuming project at once         |
 
 ### Generate an artifact on demand
 

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@erclx/canon",
   "type": "module",
-  "version": "4.110.0",
+  "version": "4.111.0",
   "description": "Infrastructure and quality tooling for developer workflows",
   "license": "MIT",
   "bin": {

package/scripts/core/capture-home.sh

@@ -0,0 +1,59 @@
+#!/usr/bin/env bash
+# Builds the landing page, captures it, and stages the frames over
+# assets/evidence/home, the baseline `canon pr evidence` compares against.
+#
+# Two workflows call it. pr-visual-checks.yml stages the frames into the runner
+# tree to diff them against the baseline, and refresh-capture-frames.yml stages
+# them to commit as the new baseline. Both need the same frames, so one script
+# keeps the two from drifting apart.
+#
+# `tooling/web/configs/e2e/screenshot.ts` is reused as is. This checkout has no
+# web/package.json, so the preview server and the cases run against the
+# root-level scripts rather than through screenshot.sh, which assumes a
+# per-project `bun run build`/`bun run preview` pair.
+set -e
+set -o pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PROJECT_ROOT="${PROJECT_ROOT:-$(cd "$SCRIPT_DIR/../.." && pwd)}"
+
+PORT=4321
+BASE_URL="http://localhost:$PORT"
+BASELINE="$PROJECT_ROOT/assets/evidence/home"
+
+cd "$PROJECT_ROOT"
+
+bun run web:build
+
+bun run web:preview &
+PREVIEW_PID=$!
+trap 'kill "$PREVIEW_PID" 2>/dev/null || true' EXIT
+
+for _ in $(seq 1 40); do
+  curl -sSf "$BASE_URL/" >/dev/null 2>&1 && break
+  sleep 0.25
+done
+curl -sSf "$BASE_URL/" >/dev/null 2>&1 || {
+  echo "Preview server not responding on port $PORT after 10s." >&2
+  exit 1
+}
+
+(cd web && SCREENSHOT_BASE_URL="$BASE_URL" bun ../tooling/web/configs/e2e/screenshot.ts)
+
+# The committed tree is cleared first rather than copied over. A copy only adds
+# and overwrites, so a section dropped from the case list would leave its
+# frames in place and a diff would see no change. Clearing first shows that
+# section as a deletion.
+rm -rf "$BASELINE"
+mkdir -p "$BASELINE"
+cp -R web/screenshots/localhost/home/. "$BASELINE/"
+
+# The sweep holds all four widths and this repository commits two, 320 and
+# 1280. The case list is the place to change if that set moves.
+find "$BASELINE" \( -name '768--*.png' -o -name '1536--*.png' \) -delete
+
+# The shared golden config names the case that emulates nothing `default`, and
+# this repository's baseline names that frame `light`.
+find "$BASELINE" -name '*--default.png' -execdir sh -c 'mv "$1" "${1%--default.png}--light.png"' _ {} \;
+
+echo "capture-home: staged frames in assets/evidence/home"

package/src/claude/cases/setup.ts

@@ -1,9 +1,8 @@
 import type { SkillCase } from '@/claude/skills-rank'
 
 /**
- * `target-setup`, `migration-*`, `canon-*`, and `create-rule`: scaffolding,
- * proposal-only migrations, and the toolkit's own reference and feedback
- * surfaces.
+ * `target-setup`, `canon-*`, and `create-rule`: scaffolding and the
+ * toolkit's own reference and feedback surfaces.
  *
  * The five `target-setup` cases are one skill read through five entry phrasings
  * rather than five rows that collapsed to a duplicate. Each names a different
@@ -36,26 +35,6 @@ export const SETUP_CASES: readonly SkillCase[] = [
       'Check that the dev server actually starts and the end-to-end suite passes against the scaffold.',
     expect: 'target-setup',
   },
-  {
-    prompt:
-      'This CLAUDE.md file has grown huge, break it apart into the tiered context model.',
-    expect: 'migration-claude-md',
-  },
-  {
-    prompt:
-      'Move the agent-flavored docs out of the docs folder and into context.',
-    expect: 'migration-context',
-  },
-  {
-    prompt:
-      'This file was replaced by a folder, help me split its content into it.',
-    expect: 'migration-superseded',
-  },
-  {
-    prompt:
-      'This project still carries a copy of the standards folder, get rid of it and point everything at the CLI.',
-    expect: 'migration-standards-drop',
-  },
   {
     prompt:
       "Before I run this sync, tell me exactly what it's going to overwrite.",

package/src/commands/sync.ts

@@ -189,7 +189,6 @@ function renderCheck(report: CheckReport): void {
   }
 
   renderUnclaimed(report)
-  renderMigrations(report)
 
   outro()
   // Scanned domains only. Tooling renders a section on every managed target, so
@@ -334,21 +333,6 @@ function describeUnclaimed(entry: UnclaimedFolder): string {
   return `${counted}, content published at ${entry.since.slice(0, 7)}`
 }
 
-/**
- * The proposal-only skills no other field reaches.
- */
-function renderMigrations(report: CheckReport): void {
-  const { migrations } = report.reverse
-  if (migrations.length === 0) return
-
-  logStep('Migrations with a case here')
-
-  for (const candidate of migrations) {
-    logWarn(candidate.reason)
-    logInfo(`Run /canon:${candidate.skill} for a proposal.`)
-  }
-}
-
 async function runSync(target: string): Promise<number> {
   intro('canon sync')
 

package/src/design/components.ts

@@ -223,6 +223,7 @@ const TEACH_CHROME: Component = {
   reads: [
     '--color-background',
     '--color-surface',
+    '--color-chrome',
     '--color-border',
     '--color-text',
     '--color-text-body',
@@ -232,7 +233,6 @@ const TEACH_CHROME: Component = {
     '--teach-sans',
     '--teach-mono',
     '--teach-measure',
-    '--teach-chrome',
     '--teach-shadow',
     '--color-teach-accent-bg',
     '--type-body-family',
@@ -247,8 +247,7 @@ const TEACH_CHROME: Component = {
   --teach-hand: 'Virgil', 'Excalifont', cursive;
   --teach-mono: var(--type-code-family);
   --teach-measure: 52rem;
-  --teach-chrome: 52rem;
-  --teach-mast-h: 4.4rem;
+  --teach-mast-h: 3.5rem;
   --teach-shadow: 0 1px 2px rgba(20, 20, 20, 0.04);
   --color-teach-accent-bg: color-mix(in srgb, var(--color-accent) 14%, var(--color-background));
 
@@ -328,6 +327,7 @@ html.sb-shut .pane { border-left: 0; }
   background: var(--color-background);
   border-bottom: 1px solid var(--color-border);
   padding: 0 1.5rem;
+  cursor: default;
 }
 
 /* The reading bar rides the masthead's own bottom edge rather than arriving as
@@ -345,37 +345,26 @@ html.sb-shut .pane { border-left: 0; }
 }
 
 /* The height is the token rather than the sum of two paddings, so the sidebar's
-   own top row lands on the same seam the bar's bottom border draws. */
+   own top row lands on the same seam the bar's bottom border draws. The row
+   spans the window rather than the reading measure: capped, the toggle and the
+   theme control floated in from the edges by an amount that changed per page. */
 .mast {
-  max-width: var(--teach-chrome);
-  margin: 0 auto;
+  max-width: none;
+  margin: 0;
   min-height: var(--teach-mast-h);
-  padding: 0;
+  padding: 0 0.9rem;
   display: flex;
   justify-content: space-between;
   align-items: center;
-  gap: 1rem;
+  gap: 0.55rem;
   font-size: var(--t5);
   color: var(--color-muted);
+  cursor: default;
 }
 
 .mast a { color: var(--color-accent); text-decoration: none; font-weight: 600; }
 .mast a:hover { text-decoration: underline; }
 
-.theme {
-  font-family: var(--teach-sans);
-  font-size: 0.75rem;
-  font-weight: 600;
-  color: var(--color-muted);
-  background: transparent;
-  border: 1px solid var(--color-border);
-  border-radius: 99px;
-  padding: 0.25rem 0.7rem;
-  cursor: pointer;
-}
-
-.theme:hover { border-color: var(--color-accent); color: var(--color-accent); }
-
 .nav {
   width: 100%;
   margin: 0;
@@ -422,7 +411,7 @@ html.sb-shut .pane { border-left: 0; }
    restates either. */
 /* A breadcrumb rather than a row of links. The gap tightens because the
    separators now carry the spacing the gap used to. */
-.mast-left { display: flex; align-items: center; gap: 0.4rem; flex-wrap: wrap; }
+.mast-left { display: flex; align-items: center; gap: 0.3rem; min-width: 0; flex-wrap: wrap; cursor: default; }
 
 /* \`--rule\` measured 1.27:1 here. It is a border token, and a border is a
    shape a reader infers rather than a glyph they resolve, so it is the wrong
@@ -430,6 +419,7 @@ html.sb-shut .pane { border-left: 0; }
 .crumb-sep {
   color: var(--color-muted);
   font-size: var(--t4);
+  margin: 0 0.1rem;
   user-select: none;
 }
 
@@ -442,27 +432,55 @@ html.sb-shut .pane { border-left: 0; }
 
 .crumb { display: inline-flex; align-items: center; }
 
-/* The label and its caret are one segment visually and two controls in fact:
-   the label navigates, the caret opens the menu. */
-.crumb-item { display: inline-flex; align-items: center; gap: 0.1rem; }
+/* The label and its caret are one component with two zones, like a split
+   button: the label navigates, the caret opens the menu, and one surface fills
+   for both so the pair reads as a single chip rather than two loose pieces.
+   The pointer covers the gap between them, which is interactive on both sides. */
+.crumb-item {
+  display: inline-flex;
+  align-items: center;
+  gap: 0;
+  border-radius: 7px;
+  padding: 0.1rem 0.15rem 0.1rem 0.35rem;
+  cursor: pointer;
+}
+
+.crumb-item:has(details.jump):hover,
+.crumb-item:has(details.jump[open]) { background: var(--color-chrome); }
+
+/* Accent reports where you are. A crumb is navigation you can take, so it is
+   neutral, and \`.mast a\` keeps the accent for any other bar link. */
+.mast a.crumb { color: var(--color-text-secondary); padding: 0.1rem 0.15rem 0.1rem 0; }
+.mast a.crumb:hover { text-decoration: none; }
+.crumb-item:hover a.crumb { color: var(--color-text); }
+
+.crumb-here { color: var(--color-text); font-weight: 650; }
 
-.crumb-here { color: var(--color-text); font-weight: 600; }
+/* \`align-items: center\` centres boxes, and a label's line box carries
+   half-leading above the cap and descender room below, so its capitals sit
+   above its own centre by an amount that scales with the font size. Trimming
+   the label to its cap band fixes the box rather than nudging the mark. It
+   lands on a span holding the text alone, since the link's own \`inline-flex\`
+   ignores it. Firefox ships no \`text-box\` and keeps the caret a pixel low. */
+.crumb-t, .crumb-here, .sb-ws .ws-name {
+  display: block;
+  text-box: trim-both cap alphabetic;
+}
 
-/* The caret is now the whole control, so it carries the 24 pixel minimum on
-   its own rather than inheriting a label's width. */
+/* The caret is the whole control, so it carries the 24 pixel minimum on its
+   own rather than inheriting a label's width. It fills nothing itself, since
+   the chip around it does. */
 .crumb-item > .jump > summary {
   min-width: 1.5rem;
   min-height: 1.5rem;
   display: inline-flex;
   align-items: center;
   justify-content: center;
-  padding: 0;
+  padding: 0.15rem 0.3rem 0.15rem 0;
   border: 0;
   background: transparent;
 }
 
-.crumb-item > .jump > summary:hover { background: var(--color-surface); border-radius: 5px; }
-
 /* A bar link is its own tap target rather than a line of text, so it carries
    the 24 pixel minimum the a11y rule sets. */
 .mast a,
@@ -478,23 +496,25 @@ html.sb-shut .pane { border-left: 0; }
   color: var(--color-muted);
 }
 
-.mast-right { display: flex; align-items: center; gap: 0.85rem; }
+.mast-right { display: flex; align-items: center; gap: 0.2rem; margin-left: auto; }
 
+/* No border, at the fold control's size and radius, so every control on the
+   bar is drawn one way. */
 .theme {
-  width: 1.9rem;
-  height: 1.9rem;
+  width: 1.75rem;
+  height: 1.75rem;
   display: inline-flex;
   align-items: center;
   justify-content: center;
   padding: 0;
-  border: 1px solid var(--color-border);
-  border-radius: 99px;
+  border: 0;
+  border-radius: 6px;
   background: transparent;
   color: var(--color-muted);
   cursor: pointer;
 }
 
-.theme:hover { border-color: var(--color-accent); color: var(--color-accent); }
+.theme:hover { color: var(--color-text); background: var(--color-chrome); }
 .theme svg { width: 0.95rem; height: 0.95rem; display: block; }
 .theme .moon { display: none; }
 :root[data-theme="dark"] .theme .sun { display: none; }
@@ -528,7 +548,7 @@ html.sb-shut .pane { border-left: 0; }
   cursor: pointer;
 }
 
-.sb-fold:hover { color: var(--color-text); background: var(--color-surface); }
+.sb-fold:hover { color: var(--color-text); background: var(--color-chrome); }
 
 .sb {
   flex: 0 0 var(--sb-w, 16rem);
@@ -564,7 +584,7 @@ html.sb-shut .sb { flex-basis: 0; width: 0; overflow: hidden; }
   display: inline-flex;
   align-items: center;
   gap: 0.35rem;
-  padding: 0.2rem 0.4rem;
+  padding: 0.25rem 0.3rem;
   border-radius: 6px;
   font-size: var(--t4);
   font-weight: 650;
@@ -572,7 +592,7 @@ html.sb-shut .sb { flex-basis: 0; width: 0; overflow: hidden; }
 }
 
 .sb-ws > summary::-webkit-details-marker { display: none; }
-.sb-ws > summary:hover, .sb-ws[open] > summary { background: var(--color-surface); color: var(--color-text); }
+.sb-ws > summary:hover, .sb-ws[open] > summary { background: var(--color-chrome); color: var(--color-text); }
 .sb-ws .car { color: var(--color-muted); display: inline-flex; align-items: center; }
 
 .sb-wl {
@@ -585,7 +605,7 @@ html.sb-shut .sb { flex-basis: 0; width: 0; overflow: hidden; }
   border: 1px solid var(--color-border);
   border-radius: 9px;
   box-shadow: 0 8px 24px rgba(0, 0, 0, 0.14);
-  padding: 0.25rem;
+  padding: 0.3rem;
 }
 
 :root[data-theme="dark"] .sb-wl { box-shadow: 0 8px 28px rgba(0, 0, 0, 0.5); }
@@ -644,7 +664,7 @@ html.sb-shut .sb { flex-basis: 0; width: 0; overflow: hidden; }
   line-height: 1.45;
 }
 
-.sb-l:hover { background: var(--color-surface); color: var(--color-text); }
+.sb-l:hover { background: var(--color-chrome); color: var(--color-text); }
 
 .sb-n {
   flex: none;
@@ -764,46 +784,47 @@ html.sb-shut .sb-grip { display: none; }
 }
 
 .jump summary::-webkit-details-marker { display: none; }
-.jump summary:hover { border-color: var(--color-accent); color: var(--color-accent); }
-/* The caret says "this is a menu" and nothing else. It used to flip to say
-   "the menu is open", which the open panel underneath already says, so the
-   motion carried no information a reader could not already see. */
-.jump summary .caret { opacity: 0.55; }
-.jump summary:hover .caret { opacity: 1; }
-.jump[open] summary { border-color: var(--color-accent); color: var(--color-accent); }
-.jump[open] summary .caret { opacity: 1; }
+/* The caret says "this is a menu" and nothing else, so it is drawn at full
+   strength in every state and stays muted. A reveal on hover hid the only
+   route into the menu from a reader who cannot hover, and an affordance is not
+   a state, so it takes no accent on hover or while open. */
+.jump summary .caret { color: var(--color-muted); }
 
 .jump-list {
   position: absolute;
   top: calc(100% + 0.4rem);
   left: 0;
   z-index: 20;
-  min-width: 21rem;
   background: var(--color-background);
   border: 1px solid var(--color-border);
   border-radius: 9px;
   box-shadow: 0 8px 24px rgba(0, 0, 0, 0.14);
-  padding: 0.35rem;
+  padding: 0.25rem;
   list-style: none;
   margin: 0;
 }
 
 :root[data-theme="dark"] .jump-list { box-shadow: 0 8px 28px rgba(0, 0, 0, 0.5); }
 
+/* One row recipe wherever the menu is mounted, on the steps the lesson list
+   beside it already reads: the label at \`--t5\` and the numeral at \`--t6\` mono.
+   The weight is restated because \`.mast a\` sets 600 for the bar's own links. */
 .jump-list a {
   display: grid;
   grid-template-columns: 1.65rem 1fr auto;
-  gap: 0.7rem;
+  gap: 0.55rem;
   align-items: center;
-  padding: 0.45rem 0.55rem;
+  padding: 0.38rem 0.55rem;
   border-radius: 6px;
   text-decoration: none;
   color: var(--color-text-body);
-  font-size: 0.875rem;
+  font-size: var(--t5);
+  font-weight: 400;
+  line-height: 1.45;
 }
 
-.jump-list a:hover { background: var(--color-surface); }
-.jump-list .n { font-family: var(--teach-mono); font-size: 0.75rem; color: var(--color-muted); }
+.jump-list a:hover { background: var(--color-chrome); }
+.jump-list .n { font-family: var(--teach-mono); font-size: var(--t6); color: var(--color-muted); }
 /* The third column used to hold a status dot that read the same on every row,
    so it reported nothing. It is a per-entry trailing slot now: the workspace
    menu puts a lesson count there and the lesson menu leaves it empty. */
@@ -825,11 +846,11 @@ html.sb-shut .sb-grip { display: none; }
 .jump-list li.all a { color: var(--color-accent); }
 .jump-list li.all .n { color: var(--color-accent); }
 
-/* The menu is \`min-width: 21rem\` and anchored to its summary, which puts its
-   right edge 35 pixels past a 390 pixel viewport. Clamping the width and
-   letting a long row wrap keeps it on screen without a breakpoint. */
+/* The menu fits its content above an 11rem floor rather than padding three
+   short titles out to a wide panel. It is anchored to its summary, so the
+   clamp keeps a long row wrapping on screen without a breakpoint. */
 .jump-list {
-  min-width: min(21rem, calc(100vw - 2rem));
+  min-width: min(11rem, calc(100vw - 2rem));
   max-width: calc(100vw - 2rem);
 }
 
@@ -850,8 +871,7 @@ html.sb-shut .sb-grip { display: none; }
 @media (max-width: 640px) {
   .bar { padding: 0 1rem; }
     main { padding-left: 1rem; padding-right: 1rem; }
-    .mast { font-size: 0.75rem; gap: 0.5rem; }
-    .mast-left { gap: 0.5rem; row-gap: 0.35rem; }
+    .mast { font-size: 0.75rem; }
 }
 
 @media (max-width: 640px) {
@@ -933,7 +953,7 @@ html.sb-shut .sb-grip { display: none; }
     cursor: pointer;
   }
 
-  .sb-close:hover { color: var(--color-text); background: var(--color-surface); }
+  .sb-close:hover { color: var(--color-text); background: var(--color-chrome); }
 
   .sb-scrim {
     display: block;

package/src/shipped/references.ts

@@ -172,8 +172,7 @@ const DOCS_PATH = /(?<![\w./-])docs\/[^\s`)\]]*\.md\b/g
  * deliberate placeholder rather than a real citation. A body illustrating
  * the shape a project's own `standards/<name>.md` takes writes a bracketed
  * token this pattern also matches, such as `standards/<slug>.md` in
- * `create-standard/SKILL.md` or `standards/<name>.md` in
- * `migration-standards-drop/SKILL.md`, and neither names a real file. A bare,
+ * `create-standard/SKILL.md`, which names no real file. A bare,
  * non-bracketed match now reports whether or not the file exists, unlike
  * `DOCS_PATH`, which still gates on existence: a same-repository citation
  * under this corpus names no target project's own tree the way a `docs/`