@design-intelligence/ghost 0.22.1 → 0.24.0

package/dist/skill-bundle/references/schema.md

@@ -1,19 +1,19 @@
 ---
 name: schema
-description: The Ghost fingerprint package shape: flat nodes, derived concreteness, Skeletons, probes, and checks.
+description: The ghost package shape: flat nodes, derived concreteness, Skeletons, probes, and checks.
 ---
 
-# Ghost Fingerprint Package Reference
+# ghost Package Reference
 
 Canonical package:
 
 ```text
 .ghost/
-  manifest.yml        ghost.fingerprint-package/v1: schema + id + optional cover
+  manifest.yml        ghost.package/v1: schema + id + optional cover
   glossary.md         kind vocabulary + meanings
   materials/          bundled materials; never a node source
-  <kind>.<slug>.md    a brand truth of a declared kind
-  <slug>.md           a brand truth without a kind
+  <kind>.<slug>.md    guidance of a declared kind
+  <slug>.md           guidance without a kind
   checks/             optional review assertions; never a node source
 ```
 
@@ -48,13 +48,17 @@ Use the full lockup when recognition matters.
 
 - Identity is the filename minus `.md`.
 - Kind is the first dotted segment of the filename.
-- `description` is the retrieval payload shown by `ghost gather`.
+- `description` is the retrieval payload shown by `ghost gather`: what the node
+  governs, the observable condition under which it applies, and what it
+  contributes where useful. Avoid broad universal wording unless universal
+  retrieval is intended.
 - `materials` accepts repo-relative paths/globs plus absolute HTTPS URLs. It is
   a locator list, not guidance.
 
-Ghost derives whether a node carries concrete material from structure: non-empty
-`materials`, a fenced code block of at least 3 lines, or a `## Skeleton` section.
-This is reported in gather/pulse and used for pull ordering.
+ghost derives whether a node carries concrete material from structure:
+non-empty `materials`, a fenced code block of at least 3 lines, or a
+`## Skeleton` section. `gather` reports these payload labels for clarity; they
+are not ranking signals.
 
 ## Skeleton convention
 
@@ -96,13 +100,13 @@ Grade whether the change preserves the logo guidance in `asset.logo`.
 `references` are node ids with optional heading anchors. `probe` is optional: a
 repo-root shell command that `ghost review` runs for offered checks by default
 (timeout 30s; stdout/stderr truncated). Probe output is evidence only, never a
-Ghost pass/fail verdict. Use `ghost review --no-probes` to skip. Trust model:
+ghost pass/fail verdict. Use `ghost review --no-probes` to skip. Trust model:
 probes are the same class as npm scripts; Git review is the boundary.
 
 ## Gather / Pull / Review
 
-- `ghost gather` emits the cover above the node menu, then coverage counts.
-  Checks are invisible.
+- `ghost gather` emits the cover above Available guidance, then coverage counts.
+  The guidance list is complete, unfiltered, and unranked. Checks are invisible.
 - `ghost pull` emits selected nodes in steering order and inlines small local
   materials. Binary local materials become inspect-pointers.
 - `ghost review` matches diff files to local node materials, offers relevant

package/dist/skill-bundle/references/self-check.md

@@ -1,30 +1,30 @@
 ---
 name: self-check
-description: A pre-generation probe that reveals whether you have actually gathered and reasoned about the brand fingerprint, before you build.
+description: A pre-generation probe that reveals whether you have actually gathered and reasoned about the brand guidance, before you build.
 ---
 
 # Recipe: Self-Check Before Generating
 
 Before writing UI, copy, email, or any output, check whether you are grounded in
-the fingerprint or about to fall back on generic instincts. These questions test
-your **grounding and provenance**, not the fingerprint's contents, so they hold
-for any fingerprint, however sparse, whatever the medium, whoever authored it.
+the guidance or about to fall back on generic instincts. These questions test
+your **grounding and provenance**, not the package's contents, so they hold
+for any ghost package, however sparse, whatever the medium, whoever authored it.
 
 Ask yourself:
 
-1. **What did you gather?** Which Ghost nodes did you pull for this task (from
+1. **What did you gather?** Which ghost nodes did you pull for this task (from
    `ghost gather`), and can you cite them by id? If you have not gathered, you
    are not grounded. Gather first.
-2. **What is Ghost-backed vs. provisional?** For each claim you are about to
+2. **What is ghost-backed vs. provisional?** For each claim you are about to
    encode, is it backed by a gathered node (cite the id), or is it your own
    provisional local reasoning? You must be able to label every claim as one or
    the other.
-3. **Do the conditions apply?** For each conditional truth you pulled, does its
+3. **Do the conditions apply?** For each conditional node you pulled, does its
    stated situation actually hold for this task? For each kind with conditional
    or scoped meaning **per the glossary**, apply it only when its stated
    situation holds; do not apply it where it does not, and do not ignore it
    where it does.
-4. **Where is the fingerprint silent?** What does the fingerprint not cover for
+4. **Where is the package silent?** What does the package not cover for
    your task, and what will carry the reasoning in those gaps? Naming the silence
    is part of being grounded; pretending coverage you do not have is not.
 
@@ -40,11 +40,11 @@ Before generating, can you cite:
 - the hard invariants?
 - the conditions that apply or do not apply?
 - any decision trace for ambiguous tradeoffs?
-- where the fingerprint is silent?
+- where the package is silent?
 
 Classify readiness:
 
-- **Green:** enough Ghost-backed guidance and concrete material for this surface
+- **Green:** enough ghost-backed guidance and concrete material for this surface
   to generate.
 - **Yellow:** generation is safe, but some reasoning must be labeled
   provisional; if there is no concrete material for this surface, readiness is
@@ -58,8 +58,8 @@ When you cannot answer the grounding questions:
    work to nodes by their descriptions.
 2. Read the selected nodes' bodies and re-ask the questions, citing node ids.
 
-A genuinely silent fingerprint is an expected state, not a blocker. When it does
+A genuinely silent package is an expected state, not a blocker. When it does
 not cover the task, say so plainly and proceed with provisional local reasoning
-when safe; label it non-Ghost-backed. If the fingerprint's cover declares a stricter
+when safe; label it non-ghost-backed. If the package's cover declares a stricter
 silence posture, honor it over this default. Ask a human before
 high-risk or brand-defining choices.

package/dist/skill-bundle/references/steering-audit.md

@@ -1,11 +1,11 @@
 ---
 name: steering-audit
-description: Audit a Ghost fingerprint for concrete steering coverage.
+description: Audit a ghost package for concrete steering coverage.
 ---
 
 # Recipe: Audit Steering Coverage
 
-A steering audit asks whether a fingerprint can move generation away from the
+A steering audit asks whether a ghost package can move generation away from the
 generic median. It is not a validation pass; `ghost validate` handles package
 shape.
 
@@ -25,31 +25,41 @@ Report first:
 
 - **Concreteness coverage:** total nodes, concrete-material nodes, prose-only
   nodes. Concrete means non-empty `materials`, a fenced code block of at least 3
-  lines, or a `## Skeleton` section.
-- **Pulse by concreteness:** concrete exposure/pull rate vs prose-only
-  exposure/pull rate. This is the tuning instrument: if concrete nodes are not
-  pulled, descriptions or task selection are failing.
+  lines, or a `## Skeleton` section. `ghost gather` also breaks out materials,
+  substantial fenced examples, and Skeletons as payload labels.
+- **Pulse by concreteness:** concrete-material exposure/pull rate vs prose-only
+  exposure/pull rate. In markdown this is the `Concrete material` row. This is
+  the tuning instrument: if concrete nodes are not pulled when applicable,
+  descriptions or task selection are failing.
 
 ## Corpus-level table
 
 | Row | Status | Evidence | Next move |
 | --- | --- | --- | --- |
-| Retrieval | strong / weak | descriptions, ids, cover | sharpen descriptions or move universal truth to the cover |
+| Retrieval | strong / weak | descriptions, ids, cover | sharpen descriptions or move universal guidance to the cover |
 | Concreteness | strong / thin | materials, fenced examples, Skeletons | add concrete locators, exemplars, or opening structures |
 | Anti-goals | present / missing / vague | `anti-goal.*`, review packet | write not-X-instead-Y replacements and material locators |
 | Consistency | clean / conflicting | concrete bodies vs rules | update stale examples; examples average with rules |
 | Stance | present / missing | cover, `principle.*` | write forced-choice principles |
 | Materials | present / missing | `materials`, inspect-pointers | point at real assets/components/tokens |
-| Exemplars | annotated / unannotated / missing | fenced samples, screenshots | say what to copy and what is incidental |
+| Exemplars | annotated / unannotated / converging / missing | fenced samples, screenshots, repeated runs | say what to copy and what is incidental; tighten applicability when unrelated tasks converge |
 | Patterns | bound-open / loose / missing | `pattern.*`, Skeletons | state applies / bound / open and add a Skeleton when opening structure matters |
 | Checks | covered / partial / missing | checks/, probes, review packet | add checks/probes for high-risk invariants |
 | Silence posture | defined / missing | cover | say when to proceed provisionally or ask |
 
+## Exemplar convergence
+
+When auditing repeated runs, compare unrelated asks that pulled the same
+exemplar. If they reproduce the same composition, section order, or decorative
+treatment without the exemplar's stated condition requiring it, the exemplar is
+steering beyond its intent. The next move is to tighten applicability, add a
+shape-appropriate exemplar, or remove the exemplar.
+
 ## Task-level readiness
 
 For a task, gather, pull, and report:
 
-- **Green:** enough Ghost-backed concrete guidance to generate.
+- **Green:** enough ghost-backed concrete guidance to generate.
 - **Yellow:** safe to generate, but some reasoning is provisional. If there is
   no concrete material for this surface, readiness is at most Yellow.
 - **Red:** missing brand-defining, high-risk, irreversible, legal, privacy, or

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@design-intelligence/ghost",
-  "version": "0.22.1",
-  "description": "Ghost CLI for the brand fingerprint: a portable steering packet of prose truths your agent reads before it builds, with advisory review after",
+  "version": "0.24.0",
+  "description": "ghost CLI for portable .ghost/ packages of brand guidance your agent reads before it builds, with advisory review after",
   "license": "Apache-2.0",
   "author": "Block, Inc.",
   "repository": {
@@ -13,7 +13,7 @@
     "url": "https://github.com/block/ghost/issues"
   },
   "keywords": [
-    "fingerprint",
+    "brand-guidance",
     "product-surface",
     "surface-composition",
     "design-system",
@@ -48,6 +48,10 @@
       "types": "./dist/ghost-core/index.d.ts",
       "import": "./dist/ghost-core/index.js"
     },
+    "./package": {
+      "types": "./dist/package.d.ts",
+      "import": "./dist/package.js"
+    },
     "./fingerprint": {
       "types": "./dist/fingerprint.d.ts",
       "import": "./dist/fingerprint.js"