@c4a/context-cli 0.7.1 → 0.7.4

package/plugins/skills/context-code-indexer/references/metrics.md

@@ -1,20 +1,77 @@
 # Profile metric revision guide
 
-Use the current CLI audit as the sole source of actual, recommended, and hard values. This guide explains how to revise a Result; it does not define thresholds or authorize changing a denominator.
+This guide names reader-facing quality problems and explains how to revise a
+Result. It does not define a second readiness check, ask the Agent to score its
+own output, or authorize changing the profile contract.
+
+## What this catalog does not prove
+
+Read this before treating a metric catalog as proof of success. These entries
+describe useful review lenses, but they do not read prose for meaning and they
+do not replace Context's deterministic validation or final Review. A Result can
+look tidy by these measures and still fail to answer a reader's question.
+
+Count entries point to shapes that should normally disappear, coverage entries
+ask whether a known inventory was handled, and ratios help find excessive
+duplication or scaffolding. They interact, so repair the reader problem rather
+than optimizing one number in isolation.
+
+One entry, `discretionary-artifacts-per-logical-unit`, is judged
+comparatively rather than against a fixed ceiling: it reacts to inflation
+relative to the selected Bundle variant. Do not look for a constant here.
+
+Context blocks only facts it can calculate from current authority, including
+workset identity, complete question planning, inventory disposition coverage,
+evidence contracts, layout and currentness. Natural-language usefulness and
+support remain authoring and final-Review responsibilities. No Agent-supplied
+metric value can weaken or satisfy those checks.
+
+## Do not satisfy one entry by breaking another
+
+These conflicts are the common cause of a second failed audit.
+
+The inventory denominator is protected. Reduce
+`narrative-enumeration-ratio` by relocating identities into the deterministic
+catalog, never by dropping them from the Result. A dropped identity converts an
+enumeration finding into an `inventory-disposition-coverage` failure.
+
+Merging is not concatenation. When resolving
+`duplicated-fact-target-ratio`, name one canonical owner and replace the other
+copies with explicit relationships. Pasting several pages together moves the
+duplication into one page and raises its enumeration instead.
+
+A deterministic catalog is not a discretionary Artifact. When enumeration
+moves into a catalog, bind that catalog to the logical unit that owns it. Do
+not create a new optional Artifact to hold it, which trades an enumeration
+finding for artifact inflation.
+
+Re-home facts before removing a target. Satisfying
+`reference-only-reader-targets` means merging an alias or ordinary reference
+into its canonical owner. Deleting the target and its facts loses evidence and
+reopens disposition coverage.
+
+Regrouping must be semantic. Satisfying `unresolved-ordinal-partitions` by
+renaming numbered batches to non-numeric labels changes nothing; the identity
+must come from a boundary a reader can name.
 
 ## inventory-disposition-coverage
 
 ### Meaning
 
-Every identity in the CLI inventory needs an owned, excluded, or unsupported disposition.
+Every identity in the CLI inventory needs an owned, excluded, or unsupported
+disposition.
 
 ### Revise
 
-Return decisions for the reported missing identities, preserving the complete inventory denominator and citing the evidence behind exclusions or unsupported cases.
+Return decisions for the reported missing identities, preserving the complete
+inventory denominator and citing the evidence behind exclusions or unsupported
+cases.
 
 ### Positive example
 
-An internal generated file is retained in the inventory and marked excluded with its generated-source evidence.
+An internal generated file is retained in the inventory and marked excluded
+with its generated-source evidence, so a reader can see it was considered and
+why it carries no reader knowledge.
 
 ### Anti-example
 
@@ -24,11 +81,13 @@ The Result omits files that did not fit the selected template.
 
 ### Meaning
 
-The same reader fact should have one canonical target rather than competing copies across Artifacts.
+The same reader fact should have one canonical target rather than competing
+copies across Artifacts.
 
 ### Revise
 
-Choose the canonical owner, replace other copies with explicit relationships, and keep evidence on the owning fact.
+Choose the canonical owner, replace other copies with explicit relationships,
+and keep evidence on the owning fact.
 
 ### Positive example
 
@@ -42,33 +101,55 @@ Several pages repeat the same behavior paragraph with different headings.
 
 ### Meaning
 
-Reader prose should explain capabilities and boundaries instead of restating a deterministic inventory as sentences.
+Reader prose should explain capabilities and boundaries instead of restating a
+deterministic inventory as sentences.
+
+The measured scope is reader prose only. A deterministic catalog Artifact that
+the Result declares as such is outside both the numerator and the denominator:
+a complete inventory is that Artifact's entire purpose, and counting it as
+narrative enumeration would penalise the Result for holding the catalog it was
+asked to produce. Count an identity once, in the prose that mentions it. An
+identity that lives only in a declared catalog is not counted at all.
 
 ### Revise
 
-Move exhaustive identities into the deterministic catalog, group related items by a reader question, and explain responsibility, differences, and handoffs.
+Move exhaustive identities into the deterministic catalog, group related items
+by a reader question, and explain responsibility, differences, and handoffs.
+
+One allowance serves every profile, because excluding declared catalogs from
+the scope already accounts for the profiles whose reader value lies in a
+complete inventory. A catalog-heavy Result is measured on its prose like any
+other, so a breach here means the prose itself restates the catalog rather than
+explaining it — never that the profile is inherently enumerative. Do not
+rewrite a catalog into prose, and do not move prose into a catalog to change
+the count: relocating an identity is legitimate only when the catalog is where
+the reader should look for it.
 
 ### Positive example
 
-A capability section explains dispatch choices and links to a complete generated route catalog.
+A capability section explains dispatch choices and links to a complete
+generated route catalog.
 
 ### Anti-example
 
-The page turns every discovered symbol into an “observed” bullet.
+The page turns every discovered symbol into an "observed" bullet.
 
 ## normalized-template-repetition-ratio
 
 ### Meaning
 
-Repeated sentence frames with only identity substitutions indicate template residue rather than evidence-specific explanation.
+Repeated sentence frames with only identity substitutions indicate template
+residue rather than evidence-specific explanation.
 
 ### Revise
 
-Merge repeated observations, describe the shared rule once, and record meaningful exceptions with their own evidence.
+Merge repeated observations, describe the shared rule once, and record
+meaningful exceptions with their own evidence.
 
 ### Positive example
 
-A family section states the common lifecycle and separately explains the exceptional member.
+A family section states the common lifecycle and separately explains the
+exceptional member.
 
 ### Anti-example
 
@@ -78,47 +159,58 @@ Every member receives the same sentence with only its name changed.
 
 ### Meaning
 
-Reader knowledge should describe stable behavior without copying implementation bodies, generated payloads, or declaration dumps.
+Reader knowledge should describe stable behavior without copying implementation
+bodies, generated payloads, or declaration dumps.
 
 ### Revise
 
-Replace copied code with an evidence-bound behavioral statement and retain only a small excerpt when it is necessary to explain a contract or failure boundary.
+Replace copied code with an evidence-bound behavioral statement and retain only
+a small excerpt when it is necessary to explain a contract or failure boundary.
 
 ### Positive example
 
-The page explains the validation and rollback sequence and links to the exact implementation locator.
+The page explains the validation and rollback sequence and links to the exact
+implementation locator.
 
 ### Anti-example
 
-The page embeds the complete function or generated type file as its main content.
+The page embeds the complete function or generated type file as its main
+content.
 
 ## reference-only-reader-targets
 
 ### Meaning
 
-A reader target requires a CLI-authorized declaration, registration, public contract, approved Subject, or Partition Subject identity.
+A reader target requires a CLI-authorized declaration, registration, public
+contract, approved Subject, or Partition Subject identity.
 
 ### Revise
 
-Merge aliases and ordinary references into their canonical owner, or remove the target when no authorized identity observation exists.
+Merge aliases and ordinary references into their canonical owner, or remove the
+target when no authorized identity observation exists.
 
 ### Positive example
 
-A re-export alias points to the canonical public target instead of creating another page.
+A re-export alias points to the canonical public target instead of creating
+another page.
 
 ### Anti-example
 
-A frequently imported helper becomes a target solely because it has many references.
+A frequently imported helper becomes a target solely because it has many
+references.
 
 ## unresolved-ordinal-partitions
 
 ### Meaning
 
-Partition identity must come from a stable semantic boundary, not traversal order or a fixed batch label.
+Partition identity must come from a stable semantic boundary, not traversal
+order or a fixed batch label.
 
 ### Revise
 
-Regroup members by capability, entrypoint, lifecycle stage, state owner, protocol boundary, or handoff; if none applies, return the protocol outcome for CLI-owned catalog fallback.
+Regroup members by capability, entrypoint, lifecycle stage, state owner,
+protocol boundary, or handoff; if none applies, return the protocol outcome for
+CLI-owned catalog fallback.
 
 ### Positive example
 
@@ -132,15 +224,18 @@ Files are divided into successive numbered batches to reduce page size.
 
 ### Meaning
 
-Optional Artifacts must answer distinct reader questions allowed by the selected Bundle variant.
+Optional Artifacts must answer distinct reader questions allowed by the
+selected Bundle variant.
 
 ### Revise
 
-Remove unsupported optional Artifacts, merge overlapping ones, or select another CLI-eligible variant when canonical facts justify it.
+Remove unsupported optional Artifacts, merge overlapping ones, or select
+another CLI-eligible variant when canonical facts justify it.
 
 ### Positive example
 
-An examples Artifact is retained because it explains a distinct usage scenario with independent evidence.
+An examples Artifact is retained because it explains a distinct usage scenario
+with independent evidence.
 
 ### Anti-example
 
@@ -150,33 +245,40 @@ The Result creates extra pages to spread the same prose across a larger Bundle.
 
 ### Meaning
 
-Every CLI-discovered example candidate needs one explicit terminal or merge decision.
+Every CLI-discovered example candidate needs one explicit terminal or merge
+decision.
 
 ### Revise
 
-Decide each reported candidate as linked, merged, documentation-only, excluded with reason, or blocked by a material request.
+Decide each reported candidate as linked, merged, documentation-only, excluded
+with reason, or blocked by a material request.
 
 ### Positive example
 
-A duplicate scenario is merged into a canonical example through its full example identity.
+A duplicate scenario is merged into a canonical example through its full
+example identity.
 
 ### Anti-example
 
-Unclear examples disappear from the Result without a disposition.
+A candidate whose purpose was unclear is left out of the Result entirely, so a
+later reader cannot tell whether it was rejected or never examined.
 
 ## example-representative-coverage
 
 ### Meaning
 
-Eligible scenarios need retained representatives that explain setup, key calls, parameters, and expected behavior.
+Eligible scenarios need retained representatives that explain setup, key calls,
+parameters, and expected behavior.
 
 ### Revise
 
-Promote or merge evidence-backed examples for the reported uncovered scenarios and close each required facet explicitly.
+Promote or merge evidence-backed examples for the reported uncovered scenarios
+and close each required facet explicitly.
 
 ### Positive example
 
-A canonical example covers the scenario and records evidence-backed not-applicable facets.
+A canonical example covers the scenario and records evidence-backed
+not-applicable facets.
 
 ### Anti-example
 
@@ -186,16 +288,20 @@ A path is listed as an example without explaining how or why it is used.
 
 ### Meaning
 
-Eligible examples should resolve to the exact public target they demonstrate whenever that target exists.
+Eligible examples should resolve to the exact public target they demonstrate
+whenever that target exists.
 
 ### Revise
 
-Resolve aliases, repair the terminal decision chain, and bind the representative to the CLI-supplied public target identity.
+Resolve aliases, repair the terminal decision chain, and bind the
+representative to the CLI-supplied public target identity.
 
 ### Positive example
 
-A scenario variant merges into a representative whose terminal decision links the canonical public target.
+A scenario variant merges into a representative whose terminal decision links
+the canonical public target.
 
 ### Anti-example
 
-An example is linked to a similarly named internal helper or only to a directory.
+An example is linked to a similarly named internal helper or only to a
+directory.

package/plugins/skills/context-code-indexer/templates/adapter-integration.md

@@ -90,8 +90,8 @@ helper separately.
 Add structured edges only for concrete registration and call paths. Keep a
 narrative locator when dynamic dispatch prevents an unambiguous edge.
 
-Return `identityGroups` when several target identities share one explained
-adapter responsibility. Return every source-backed adjacency in
+Use one canonical semantic subject when several target identities share one
+explained adapter responsibility. Return every source-backed adjacency in
 `chainCandidates`, then provide one `chainCandidateDecisions` record for each
 candidate. A documented decision names the reader-facing view and emits its
 structured edge; an equivalent candidate merges into that canonical candidate;

package/plugins/skills/context-code-indexer/templates/component-library.md

@@ -30,6 +30,14 @@ Treat stories and examples as usage evidence, not as authority for unsupported
 props or behavior. When generated declarations disagree with source, identify
 the authoritative source and record the gap.
 
+When the Authorized Workset View supplies a public declaration together with
+structured props, events, defaults, variants, maintained examples, or source
+roles, the corresponding reader section must consume those facts or record an
+explicit supported omission/material-gap disposition. Do not replace available
+contract facts with a generic statement about the component directory. A group
+without any reader-authorizing public declaration remains catalog-only and does
+not produce an empty component page.
+
 ## Questions the knowledge must answer
 
 1. Which component families are public, and how are they imported?

package/plugins/skills/context-code-indexer/templates/gateway-facade.md

@@ -36,6 +36,22 @@ fields, imports, or transport-library dependencies.
 4. What timeout, retry, compatibility, and failure behavior is source-backed?
 5. Which generated artifacts are locators rather than independent authority?
 
+## Chapter blueprint
+
+```markdown
+# <Gateway or facade boundary>
+## Exposed and translated responsibilities
+## Operations provided or consumed
+## Request, identity, and context mapping
+## Timeout, retry, compatibility, and failure translation
+## Contract authority and generated bindings
+## Source-backed provider-to-consumer relationships
+```
+
+Operations belong under the operations chapter as canonical records. Do not
+promote a single operation to a page heading unless its authority, security, or
+failure semantics differ materially from the rest of the boundary.
+
 ## Canonical operation record
 
 Use this record wherever another selected template asks for an operation,

package/plugins/skills/context-markdown-indexer/SKILL.md

@@ -1,10 +1,10 @@
 ---
 name: context-markdown-indexer
-description: Context-managed Provider for evidence-bound document knowledge. Use only when the Context Indexer lifecycle selects this Provider, not as a standalone workflow.
+description: Context-managed Provider for source-grounded document knowledge. Use only when the Context Indexer lifecycle selects this Provider, not as a standalone workflow.
 metadata:
   context-role: "indexer-provider"
   context-public-entry: "false"
-  context-provider-version: "1.0.0"
+  context-provider-version: "1.1.0"
 ---
 
 # Context Markdown Indexer
@@ -15,6 +15,16 @@ the shared Provider/customization guide before publishing or extending this
 Skill. Those documents are authoring contracts; the current captured evidence,
 workset and Route remain execution authority.
 
-Use this Provider only through the Context Indexer lifecycle. Context supplies exact captured sources, profile authority, worksets, and evidence views; the Provider does not read arbitrary workspace paths or write approved knowledge directly.
+Use this Provider only through the Context Indexer lifecycle. Context supplies exact profile/workset authority and one Authorized Workset View containing the permitted captured evidence. The Provider does not open source-specific readers, manage pagination/receipts, read arbitrary workspace paths, or write approved knowledge directly.
+
+Treat the captured document as reader material, not as pipeline metadata. Every
+reader Artifact must preserve or synthesize the source-backed guidance needed by
+the selected profile. A heading inventory, directory summary, generic document
+description, internal Fact/View/digest count, or a sentence sending the reader
+back to the source is not useful knowledge and must not be emitted. Keep
+execution receipts, content digests, internal refs, and other recovery machinery
+out of reader Markdown. Bind each reader claim only to the smallest relevant
+source span or captured evidence set; never attach a whole document corpus or
+Authorized Workset View to a generic summary.
 
 The machine-readable authority is `context-indexer.yaml`. Context materializes the detailed guidance only after verifying this Bundle's release identity and complete file ledger.

package/plugins/skills/context-markdown-indexer/context-indexer.yaml

@@ -1,6 +1,6 @@
 protocol: context.indexer.provider/v1
 id: context-markdown-indexer
-version: 1.0.0
+version: 1.1.0
 domains: [markdown]
 
 activation:
@@ -34,16 +34,28 @@ provides:
     - test-validation
     - release-migration-guide
     - documentation-site
+  partition_strategies:
+    - id: reader-subject
+      profiles:
+        - domain-reference
+        - product-requirements
+        - technical-guide
+        - user-and-developer-guide
+        - public-api-reference
+        - runbook
+        - faq-support
+        - standard-policy
+        - decision-record
+        - incident-review
+        - test-validation
+        - release-migration-guide
+        - documentation-site
+      priority: 100
   operations:
     - id: main-index
-      consumes: context.indexer.main-workset/v1
+      consumes: context.indexer.main-workset/v2
       produces: context.indexer.main-result/v1
       accepts_layer_fragments: [fact-enrichment, template-variables]
-    - id: material-answer
-      consumes: context.indexer.material-question-workset/v1
-      produces: context.indexer.material-answer-result/v1
-      supported_evidence_kinds: [contract, configuration, documentation, runbook, decision-record, test-result, runtime-observation, tool-snapshot]
-      accepts_layer_fragments: [fact-enrichment]
   source_roles: [authoritative-document, supporting-document, operational-document, decision-record]
   logical_units:
     - id: reader-subject
@@ -145,3 +157,4 @@ quality_guidance:
     - reference-only-reader-targets
     - unresolved-ordinal-partitions
     - discretionary-artifacts-per-logical-unit
+  repair: references/indexer.md

package/plugins/skills/context-markdown-indexer/references/indexer.md

@@ -1,6 +1,6 @@
 # Markdown Indexer authoring contract
 
-Classify and author at the source-backed Section level. A document profile describes reader intent; it is not an output directory. Mixed documents may contain Sections with different reader goals, but every Section must retain one stable subject, owner, source authority, and projection intent.
+Classify and author at the source-backed Section level using only the current Authorized Workset View. A document profile describes reader intent; it is not an output directory. Mixed documents may contain Sections with different reader goals, but every Section must retain one stable subject, owner, source authority, and projection intent.
 
 Use `classification.md` as the semantic source for `document_kind`, `reader_goal`, and current `artifact_kind` selection. Use `structure-and-artifacts.md` for mixed-document routing, Section-versus-Artifact promotion, density, duplicate/conflict handling, and target candidate resolution. The CLI-supplied profile contract remains the only authority for collection and path projection.
 
@@ -14,10 +14,39 @@ Use `editorial-policy.md` for scenario editorial signals, recommended outcomes,
 
 Preserve exact commands, identifiers, links, attachments, tables, and code blocks when they are evidence-bearing. Separate deterministic catalogs from explanation. Do not publish unsupported claims, conversion annotations, placeholders, or unresolved requests as knowledge.
 
-When current source material cannot answer a required canonical question, return the exact material-question disposition for the supplied target. Answers may use only authorized evidence kinds and spans. Context owns review, layout, collection mapping, material-gap persistence, quality thresholds, and final close.
+When a captured document already contains coherent reader-facing guidance, preserve and organize that supported content instead of replacing it with a heading inventory, directory summary, or a sentence that points the reader back to the source. Summarize only repetition or navigation scaffolding. A shorter Artifact is acceptable only when it still answers the selected reader questions and retains the source's useful contracts, examples, conditions, compatibility notes, and uncertainty.
 
-Follow the operation in the supplied run request. For `main-index`, return only the current `IndexerResult`/`ArtifactResult` contract; Context derives the exact source-span, selected-fact, logical-unit, Artifact, and negative group-input dependencies from that result. For `material-answer`, return only `context.indexer.material-answer-result/v1`. Each binding must echo one eligible `question_key`, its exact `question_revision_digest`, and canonical evidence claims containing only `kind`, authorized `source_ref`, normalized `source_spans`, and the current content `evidence_digest`. Do not return reader prose, an answer body, source-origin/input identities, an `EvidenceItemRef`, a future Artifact/Section identity, or an actualization. Context canonicalizes evidence, performs the limited Review, derives a body-free planned answer from the approved binding, and maps its supplied answer landing after layout.
+For each reader Artifact, start the first actual Section with exactly one
+level-one heading that names the reader subject. Keep the heading concise and
+source-backed. It is a display title for outline and final Candidate Review,
+not a SubjectKey, ownership signal, or identity fallback. Later Sections in the
+same Artifact must not add another level-one heading.
 
-Never create a legacy `MarkdownCollectionSlice` or invoke the independent `alignProse` phase. Never mix `main-index` and `material-answer` output.
+When current source material cannot answer a required canonical question, return the exact material-question disposition for the supplied target. A later run may consume newly captured Markdown as ordinary source and update the same knowledge candidate. Context owns layout, collection mapping, runtime material-gap state, quality thresholds, the single final content Review, and close.
+
+Return only the current `main-index` `IndexerResult`/`ArtifactResult` contract. Context derives the source dependencies needed for stale detection and recovery from that result. Do not emit a separate answer body, answer-only result, future Artifact/Section landing, or post-layout actualization.
+
+Return all knowledge through the current Indexer result; do not create an independent authoring pipeline.
+
+## Quality guidance and production checks
+
+Context blocks only conditions it can derive from current inputs: the complete
+source inventory needs dispositions, the selected reader-question set needs a
+complete plan, answered questions must satisfy their evidence contracts, and
+references, layout and currentness must remain valid.
+
+Profile metrics are writing guidance rather than a separate pass decision.
+They point out duplicated facts, enumeration where explanation was expected,
+repeated template scaffolding, oversized quoted bodies, traversal-ordered
+Partitions, reference-only targets and unsupported optional Artifacts. Treat a
+finding as a prompt to reread the Section. Do not adjust wording, headings,
+sentence counts or Partitions merely to move a counter.
+
+Natural-language statements are not mechanically checked sentence by sentence.
+Use only behavior stated by the supplied material, record a material gap when
+the source does not establish a needed conclusion, and rely on final Review to
+reject unsupported interpretation. When shape guidance conflicts with source
+coverage, preserve the source-backed content and improve its presentation
+rather than deleting it.
 
 Do not return output paths, collection names, arbitrary question text, new authority, or pass/fail claims.

package/plugins/skills/context-markdown-indexer/references/semantic-planning.md

@@ -1,15 +1,31 @@
-# Evidence, subject, and claim planning
-
-Use these rules while producing the operation selected by the current run
-request. For the current `main-index` `ArtifactResult`, apply every rule below. For
-`material-answer`, produce only evidence claims for eligible questions; the
-question identity, landing, canonical evidence items, planned-answer projection
-and post-layout actualization remain Context-owned. These rules preserve the
-useful semantic judgments from the former classification and align path without
-restoring `context.structure.v1`, collection selection, or a second write
-workflow. Context remains the authority for schemas, SubjectKey normalization,
-identities, paths, collisions, stale state, layout changes, Review, and
-publication.
+# Source, subject, and claim planning
+
+Use these rules while producing the current `main-index` `ArtifactResult`.
+Captured Markdown is ordinary source material: it may create or enrich the same
+knowledge candidates as code-derived input. If required material is still
+missing, return a material-gap disposition and let a later main-index run consume
+the newly captured source. Do not create an answer-only result, landing, or
+second review workflow. Context remains the authority for schemas, SubjectKey
+normalization, identities, paths, collisions, stale state, layout changes,
+Review, and publication.
+
+## `reader-subject` partition strategy
+
+For a partition workset, group captured document members by durable reader
+subject rather than by file, heading, route, or temporary capture batch. Use
+the complete authorized document text together with maintained title,
+`source_path`, route, audience, and reader-task evidence. These fields are
+evidence for the decision; none is sufficient by itself.
+
+One group may contain multiple documents when they jointly explain the same
+reader subject. Split platform or runtime variants only when their supported
+contract, behavior, lifecycle, or reader task is materially different. Keep
+navigation-only indexes, generated duplicates, empty placeholders, and
+superseded pages out of authored groups with an explicit inventory
+disposition. Every current inventory member must still receive exactly one
+partition disposition. If durable subject boundaries cannot be established,
+fail this semantic strategy so Context can use its existing catalog fallback;
+do not silently return one group per file under `reader-subject`.
 
 ## Evidence and authority
 
@@ -18,8 +34,8 @@ making a source-wide decision. A file name, URL, title, heading, navigation
 label, polished wording, profile id, or example is a navigation signal, not
 proof of subject identity, authority, or reader intent.
 
-Use only current source roles, evidence bindings, target-resolution views, and
-question targets supplied with the workset. Supporting or context-only material
+Use only current source roles, authorized source views, target-resolution views,
+and question targets supplied with the workset. Supporting or context-only material
 may guide investigation but cannot become a cited Section or structured claim
 unless the current authority makes it eligible. Preserve exact identifiers,
 commands, links, numbers, conditions, code, and source-authored uncertainty.

package/plugins/skills/context-markdown-indexer/references/structure-and-artifacts.md

@@ -32,6 +32,12 @@ or a long table. Do not create thin Artifacts to improve counts. If material has
 an independent subject rather than merely a separate reader page, use the
 workset's target-resolution choices; never invent a Node identity from a title.
 
+Do not collapse a substantial captured document into a page that only lists its
+headings or says the source should be consulted. Preserve supported procedures,
+API tables, examples, constraints, compatibility notes, and decisions inside
+the selected reader Artifact; remove only navigation chrome, conversion noise,
+duplicates, and content with an explicit omission disposition.
+
 Within one Artifact, every Section must use that Artifact's `artifact_kind`.
 Start by grouping only Sections with the same exact
 `document_kind + reader_goal + artifact_kind` intent. Different intents may be

package/plugins/skills/context-markdown-indexer/tests/fixtures/anonymous.json

@@ -18,7 +18,7 @@
       "artifact_kind": "content",
       "purpose": "required",
       "reader_question_refs": [
-        "question:source-authority"
+        "question:reader-structure"
       ],
       "evidence_refs": [
         "evidence:anonymous-guide-source"