@c4a/context-cli 0.7.26 → 0.7.30

package/plugins/skills/context-inspect-search/SKILL.md

@@ -25,14 +25,37 @@ Do not execute returned production actions, initialize a replacement workspace,
 or install tools without existing authorization. Querying must not alter source
 registrations, snapshots, tasks, approvals or user checkouts.
 
-## Search available material concurrently
-
-Check relevant knowledge packages, the workspace and reusable source checkouts
-in parallel. Search each as soon as it is readable; tool checks and authorized
-upgrades must not block independent reading. If a missing configured workspace
-is needed for further attribution, recover it into an isolated directory while
-package retrieval continues. Without a configured workspace, investigate the
-available packages directly; do not request or create one merely to start.
+Resolve `CONTEXT_QUERY_SOURCE_MODE` from the host or Bot instance configuration
+before choosing the first retrieval path. Supported values are:
+
+- `repo-first` (default): search approved `knowledge/` in the configured knowledge
+  workspace first; trace authorized original sources only for material gaps.
+  Use an available package when approved knowledge is unavailable.
+- `package-first`: inspect the selected package first, then use its recorded
+  workspace or repository sources for missing, stale or disputed evidence.
+- `dual`: inspect the package and repository paths concurrently, then reconcile
+  relevant differences before answering.
+
+An unknown or empty value is `repo-first`; do not invent a fourth mode. The
+mode changes retrieval order, not authorization or evidence standards. It does
+not make a missing secondary source mandatory: continue with readable configured
+sources and identify only gaps that affect the answer.
+
+## Search available material according to the configured mode
+
+Apply `CONTEXT_QUERY_SOURCE_MODE` before starting retrieval. In `repo-first`,
+start with approved knowledge, not original-code checkout preparation. With
+readable `knowledge/`, do not probe `dist/`, build inventories or installed
+packages: distribution output is not a query prerequisite. An explicit request
+to investigate delivered content is an exception. In `package-first`, start with
+the selected readable package. In `dual`, start both paths concurrently. Search
+each selected path as soon as it is readable; tool checks and authorized
+upgrades must not block independent reading. If a configured primary path is
+missing, continue with the readable fallback instead of stopping. If a missing
+configured workspace is needed for further attribution, recover it into an
+isolated directory while other selected retrieval continues. Without a
+configured workspace, investigate available packages directly; do not request
+or create one merely to start.
 
 A knowledge package may be workspace build output (usually `dist/`), a global
 installation, or an equivalent host-provided directory. Locate it from available
@@ -44,9 +67,11 @@ missing, use the host's configured, authorized package retrieval mechanism while
 other reading continues. Only check or prepare tools needed for the next actual
 operation; installation failure does not block independent local retrieval.
 
-For gaps or latest-content questions, also search approved knowledge when a workspace is available:
-use `knowledge/structure.yaml` and relevant `knowledge/` paths, trying business
-terms, symbols and synonyms, then read matching sections in context. Delivered
+Search relevant `knowledge/` paths using business terms, symbols and synonyms,
+then read matching sections in context; use `knowledge/structure.yaml` when
+navigation or associations are needed, not as an exhaustive preflight. In
+package-first or dual mode, approved knowledge can resolve gaps or freshness
+questions in delivered material. Delivered
 content may lag approved knowledge; they are not independent corroboration.
 Candidates and temporary reports are not approved pages. A search miss does not
 prove a capability is absent. Do not build merely to answer a query; a separately
@@ -56,8 +81,11 @@ When package or approved knowledge leaves a gap, directly trace relevant sources
 without asking whether to deepen the query. Use available document, code-search
 or extraction tools within the authorized scope, recording source identity and
 version. Prepare sources while continuing other reading.
-One failed source blocks only dependent claims. Answer when evidence is
-sufficient, without completing unnecessary preparation.
+One failed source blocks only dependent claims. If the read knowledge supports
+the requested conclusion, its conditions and scope, answer without restoring
+code or checking every associated source. Missing mechanisms, conflicting facts
+or unsupported current-behavior claims still require targeted investigation;
+an article title, search snippet or unread citation is not sufficient evidence.
 
 ## Trace and retrieve only relevant sources
 
@@ -68,18 +96,23 @@ registry is not a mandatory intermediate step. If the source identity or version
 cannot be established, retain package-grounded findings and qualify attribution;
 do not invent a repository or equate current source with the package's baseline.
 
-When a workspace and `context-build-inventory.json` are available, map a delivered page's
+Only when tracing a delivered page and a build inventory is already available, map its
 `dist_path` through `approved_knowledge.files` to its `approved_path`, relative
-to the workspace's `knowledge/` root. Follow the approved page's associations in
-`knowledge/structure.yaml` and `sources/*/index.yaml`. Without a reliable mapping,
+to the workspace's `knowledge/` root. For directly read approved pages, skip this
+mapping entirely. Follow relevant associations in `knowledge/structure.yaml`
+and `sources/*/index.yaml` only when source tracing is needed. Without a reliable mapping,
 continue package retrieval and follow explicit source references where available.
 Do not guess originals from similar filenames or claim attribution without
 checking the referenced material.
 
 Use the repository and recorded commit identified by the package or workspace
 as the source baseline.
-Reuse existing checkouts only after checking remote, commit, module coverage
-and local changes; do not reset them or change their sparse configuration.
+Before reusing a checkout for source reads, group checks of remote, commit,
+module coverage and local changes. Reuse those findings within this response
+instead of checking again before each search or file read. Recheck affected
+facts if the checkout/ref changes, sparse scope expands or there is evidence of
+concurrent edits; do not treat a previous response's checks as current. Do not
+reset user checkouts or change their sparse configuration.
 Missing code goes into a reusable query-owned path such as
 `.tmp/context-inspect/<host>/<repository-path>/<commit>/`, separate from user
 checkouts and production-managed sources. Use safe, credential-free path
@@ -108,17 +141,45 @@ capture/recovery actions. Production recovery gates do not block this independen
 inspection. Preserve unavailable-evidence gaps rather than changing production
 state to get past them.
 
-Compare relevant baseline files with current code when accessible, including
-relevant local changes. Use actual diffs to assess impact; current code is not a
-substitute for the baseline. Mention a missing comparison only when it limits
-the answer. For documents, notes and sessions, read saved bodies and necessary
+Batch independent searches and related file reads once paths are known, such as
+the entrypoint, state update and controlling render logic for one mechanism.
+Read enough surrounding context and follow unresolved imports or calls; batching
+must not truncate decisive evidence or turn a partial read into a complete audit.
+Reuse already read evidence rather than rereading it to prepare each tool call.
+
+Let evidence sufficiency determine whether a version comparison is needed.
+When the available knowledge or identified baseline code resolves the question,
+answer without fetching the default branch, recovering a second checkout or
+reading history merely to confirm freshness. Compare narrowly when uncertainty,
+conflicting evidence, suspected changes or an explicit version-comparison request
+makes it useful. A question about current behavior requires judging whether the
+available evidence supports that claim, not automatically comparing every branch.
+Retain necessary source reads; do not equate a recorded baseline or branch HEAD
+with a verified live deployment. When comparing, use actual relevant diffs,
+including relevant local changes; current code is not a substitute for the baseline.
+Mention a missing comparison only when it limits the answer.
+For documents, notes and sessions, read saved bodies and necessary
 attachments; fetch missing accessible material within scope. A summary is not a
 full transcript, an unread link is not evidence, and current remote content does
 not prove a historical snapshot.
 
 ## Answer with evidence and hand off updates
 
-Lead with the conclusion and a concise evidence chain. Distinguish implementation,
+Answer the user's question directly, keeping conditions that change the conclusion
+beside it. For complex questions, explain the mechanism that determines the result,
+not just the verdict and source links. Explain its practical meaning before using
+a short code excerpt or concrete rule when that helps the reader verify or act.
+Keep decisive evidence beside the claim it supports; separate supplementary facts
+from the proof. State where a local rule applies rather than generalizing it to
+the whole system, and do not infer presentation or behavior from data retrieval
+alone without checking the controlling logic.
+
+Adapt depth and structure to the question: keep simple answers short; use short
+headings and, where supported, separators between major semantic sections in
+longer answers. Do not impose a fixed outline, separate every paragraph, repeat
+the conclusion, or remove necessary explanation just to shorten the answer.
+Keep conditions affecting the answer prominent and supplementary reading secondary.
+Distinguish implementation,
 declared contracts, test assertions, runtime observations and inferences. Perform
 bounded non-destructive validation only when useful; inspect its effects first
 and do not run unrelated scripts, builds or tests.
@@ -130,13 +191,25 @@ limitation. Do not dump internal reasoning, runtime identifiers or routine
 version comparisons. Explain only differences affecting the answer; hashes may
 appear in source URLs without requiring a separate version audit in the prose.
 
-For website links, use `context-site-map.json` from the selected package or
-sibling website output. Match `pages[].package_path` or `approved_path`, and
+Collect citation targets after the relevant evidence is established and resolve
+them in one batch using the host's existing link resolver when available. Reuse
+known workspace, repository and site parameters; do not rediscover them per link.
+If later evidence adds targets, resolve only those missing from the results.
+Without a host resolver, use an already available `context-site-map.json` from
+the selected package or website output. Do not search for `dist/` merely to
+obtain links for readable approved knowledge. Match `pages[].package_path` or `approved_path`, and
 resolve `site_path` against `site_url` without prepending `base` again. Cite only
 matched pages, deduplicate links and do not invent anchors. If mapping is absent
 or invalid, retain local citations. Do not build, publish or probe remote sites
 just to format citations; a local map is not proof of the current online content.
 
+Once the requested questions are supported or genuinely unavailable evidence is
+clearly bounded, synthesize the final answer directly. Do not add another planning,
+freshness-audit or summary round just to prepare it. Preserve necessary mechanisms,
+exceptions, citations and verification limits; fewer preparation steps must not
+mean weaker evidence or omitted questions. Host delivery and receipt requirements
+still apply, without repeating the completed investigation.
+
 Suggest a knowledge update only with supporting evidence: identify the affected
 page or gap, proposed change and bounded source scope. Check neighboring content
 before proposing a new page. Only after user acceptance, hand the evidence and

package/providers/context/graphs/indexer.yaml

@@ -36,6 +36,10 @@ nodes:
     kind: action
     reasonCode: route.production.planning-required
     action: actions/work-production-stage.yaml
+    resources:
+      recommended:
+        - resources/procedures/document-code-investigation.md
+        - resources/procedures/repository-source-recovery.md
     satisfiedBy:
       - path: production.planning_complete
         equals: true

package/providers/context/manifest.json

@@ -2,14 +2,14 @@
   "schema": "agent-graph.bundle.v1",
   "provider": {
     "id": "c4a/context",
-    "version": "0.7.26"
+    "version": "0.7.30"
   },
   "providerManifest": "provider.yaml",
   "graphs": [
     {
       "id": "indexer",
       "path": "graphs/indexer.yaml",
-      "digest": "sha256:4992b548a9d19c6d7fd25a5dc0ae6b78a4a494ce39d37633144d389362f21122"
+      "digest": "sha256:d52242180fbe1afea09946bb223a698f7e0637f81d619d9245acb74732f2409e"
     },
     {
       "id": "workspace",
@@ -183,7 +183,7 @@
     {
       "id": "context.sdk.lark-resources",
       "path": "resources/manuals/guides/lark-resources.md",
-      "digest": "sha256:b1ffb74a145962ff89a5db29cffd1318e4ec038b97884dcd43804170cb0f6f64"
+      "digest": "sha256:d5b6b44fe800f302232300adbdb4f32b04b78588ea7879dc9a5c247f36658b9d"
     },
     {
       "id": "context.sdk.package-outputs",
@@ -248,7 +248,7 @@
     {
       "id": "dialogue.document-capture",
       "path": "resources/dialogue/document-capture.md",
-      "digest": "sha256:eab34072dc96d77c0749148463f722d2483d8e7d83899ce489e927176a1178e4"
+      "digest": "sha256:20bb708f2c471591c0f633742d75a1353bcd4762917ce9528d3de0c7f4babbe8"
     },
     {
       "id": "dialogue.human-gates",
@@ -288,7 +288,12 @@
     {
       "id": "procedure.document-capture",
       "path": "resources/procedures/document-capture.md",
-      "digest": "sha256:9d0d1b9187bbe540a4afffb366cd549385dbf61bfaffa96ae51de848b2dd5460"
+      "digest": "sha256:2e6b4e45587fffc1930a3dce0fc6b68d659499a034a367e1cc34a2ef46cb50d9"
+    },
+    {
+      "id": "procedure.document-code-investigation",
+      "path": "resources/procedures/document-code-investigation.md",
+      "digest": "sha256:eff8e54b9e1c1139c5df82a0685a365bce902dfb4a19c17e380967ee1c1762f5"
     },
     {
       "id": "procedure.homogeneous-source-review",
@@ -343,7 +348,7 @@
     {
       "id": "procedure.repository-source-recovery",
       "path": "resources/procedures/repository-source-recovery.md",
-      "digest": "sha256:024f6e2947ef421b9a5f5e138c9af65a054d676a7f484dbf80b43a620ee2c95d"
+      "digest": "sha256:20a8112df985a20481e914a564c64c231218c73c1630d74cd46d72987e54f714"
     },
     {
       "id": "procedure.runtime-event-delivery",
@@ -363,7 +368,7 @@
     {
       "id": "procedure.source-capture-detailed",
       "path": "resources/procedures/source-capture-detailed.md",
-      "digest": "sha256:75694361543f5c44a1077b5e6c4ed9bca844ec9b46c1773198f66e0f20ba3748"
+      "digest": "sha256:39644875f09bac66580332357c3d9701c234da036b61205c8a02029165455f76"
     },
     {
       "id": "procedure.source-update",
@@ -562,7 +567,7 @@
     },
     {
       "path": "graphs/indexer.yaml",
-      "digest": "sha256:4992b548a9d19c6d7fd25a5dc0ae6b78a4a494ce39d37633144d389362f21122"
+      "digest": "sha256:d52242180fbe1afea09946bb223a698f7e0637f81d619d9245acb74732f2409e"
     },
     {
       "path": "graphs/workspace.yaml",
@@ -570,7 +575,7 @@
     },
     {
       "path": "provider.yaml",
-      "digest": "sha256:7e5a25e5636983a6fb881621e826c102680f44cde9d21086c627861e119d1f7a"
+      "digest": "sha256:50adfb493675f902e6df4d6eade84bb279c64b767084aa15277d65870825458e"
     },
     {
       "path": "resources/contracts/indexer-provider-guide.yaml",
@@ -594,7 +599,7 @@
     },
     {
       "path": "resources/dialogue/document-capture.md",
-      "digest": "sha256:eab34072dc96d77c0749148463f722d2483d8e7d83899ce489e927176a1178e4"
+      "digest": "sha256:20bb708f2c471591c0f633742d75a1353bcd4762917ce9528d3de0c7f4babbe8"
     },
     {
       "path": "resources/dialogue/human-gates.md",
@@ -630,7 +635,7 @@
     },
     {
       "path": "resources/manuals/guides/lark-resources.md",
-      "digest": "sha256:b1ffb74a145962ff89a5db29cffd1318e4ec038b97884dcd43804170cb0f6f64"
+      "digest": "sha256:d5b6b44fe800f302232300adbdb4f32b04b78588ea7879dc9a5c247f36658b9d"
     },
     {
       "path": "resources/manuals/guides/markdown-indexer-skill-authoring.md",
@@ -690,7 +695,11 @@
     },
     {
       "path": "resources/procedures/document-capture.md",
-      "digest": "sha256:9d0d1b9187bbe540a4afffb366cd549385dbf61bfaffa96ae51de848b2dd5460"
+      "digest": "sha256:2e6b4e45587fffc1930a3dce0fc6b68d659499a034a367e1cc34a2ef46cb50d9"
+    },
+    {
+      "path": "resources/procedures/document-code-investigation.md",
+      "digest": "sha256:eff8e54b9e1c1139c5df82a0685a365bce902dfb4a19c17e380967ee1c1762f5"
     },
     {
       "path": "resources/procedures/homogeneous-source-review.md",
@@ -734,7 +743,7 @@
     },
     {
       "path": "resources/procedures/repository-source-recovery.md",
-      "digest": "sha256:024f6e2947ef421b9a5f5e138c9af65a054d676a7f484dbf80b43a620ee2c95d"
+      "digest": "sha256:20a8112df985a20481e914a564c64c231218c73c1630d74cd46d72987e54f714"
     },
     {
       "path": "resources/procedures/runtime-event-delivery.md",
@@ -750,7 +759,7 @@
     },
     {
       "path": "resources/procedures/source-capture-detailed.md",
-      "digest": "sha256:75694361543f5c44a1077b5e6c4ed9bca844ec9b46c1773198f66e0f20ba3748"
+      "digest": "sha256:39644875f09bac66580332357c3d9701c234da036b61205c8a02029165455f76"
     },
     {
       "path": "resources/procedures/source-update.md",
@@ -862,12 +871,12 @@
     },
     {
       "path": "skills/work-production-stage/SKILL.md",
-      "digest": "sha256:b7aa28c032e00f93f67d3c9dc3589cf1513451456f589262e22ad7f73d265de7"
+      "digest": "sha256:676289693df57b5887f104d751dd9c8cc73fb1c903e4df7088f110ef6147388a"
     }
   ],
   "graphDependencies": {
     "indexer": [],
     "workspace": []
   },
-  "digest": "sha256:3f380c5b4a1adbc92847189b7072ec7515990dce621785bf25469268f9bbb252"
+  "digest": "sha256:36c06f801f991b92bb78cc94676c8d338389bdc031f49577eb196a96628b2f76"
 }

package/providers/context/provider.yaml

@@ -1,6 +1,6 @@
 schema: agent-graph.provider.v1
 id: c4a/context
-version: 0.7.26
+version: 0.7.30
 name: Context workflow
 description: Internal work contract for Context knowledge workspaces.
 graphs:

package/providers/context/resources/dialogue/document-capture.md

@@ -23,6 +23,14 @@ refresh, or capture preview before permission is clear. A request that
 explicitly says to ingest, fetch, capture, or read the named documents grants
 that scope; merely mentioning a possible source does not.
 
+The image preflight in the document-capture procedure is an Agent self-check,
+not an additional user question. Before capture, briefly report the task-wide
+document count and applied policy. Reference-only images keep source references
+but are not downloaded or interpreted. Do not wait for confirmation or infer an
+all-image override from managed execution. Unknown scope requires authorized
+metadata or body-and-reference-only discovery first. Existing source-read
+permission requirements remain unchanged.
+
 If deterministic inspection detects `.mdx`, route metadata, sidebars, or a docs
 configuration, explain that the boundary looks like a documentation site and
 ask whether to use the Context document-site processor. Do not infer that

package/providers/context/resources/manuals/guides/lark-resources.md

@@ -133,8 +133,24 @@ assess the page with its non-text evidence.
 
 ## Image choices for a production task
 
+Before capture, the Agent follows the document-capture Route procedure's task-wide
+document threshold and image self-check, without asking for user confirmation.
+Set `DOCUMENT_IMAGE_CAPTURE` in task instructions or workspace guidance to `ON`
+(include images), `OFF` (references only), or a positive integer document threshold.
+Omitting it uses `10`; exactly the threshold remains within normal acquisition.
+Current-conversation instructions override workspace guidance for that task only.
+For example, `DOCUMENT_IMAGE_CAPTURE: 20` skips image downloads when the task
+contains more than 20 distinct documents. This is an Agent instruction, not an
+environment variable, CLI flag or SDK field; no Bot configuration is required.
+Above a numeric threshold, the default is body capture with image references,
+unless the effective setting is overridden to `ON`. This workflow
+choice uses the existing resource fields; it does not change SDK defaults or add
+a CLI hard gate. Apply it before bulk capture or host prefetch, not after images
+have already been downloaded.
+
 The work-start report asks once when a task contains more than 30 distinct images
-and has no explicit image policy. The choices are intelligent conversion and
+that will be acquired and has no explicit image handling policy. Reuse the
+current-conversation image instruction when it already settles this choice. The choices are intelligent conversion and
 inclusion (recommended), include all as images with compression, or include none
 with placeholders. Count the whole task, not each capture or writing batch.
 Review reports show the resulting image handling and any fallbacks.

package/providers/context/resources/procedures/document-capture.md

@@ -13,14 +13,16 @@ Before capture:
 
 1. every registered document module must have a matching capture declaration;
 2. the user must have allowed the external read in the current conversation;
-3. execute only the current route command.
+3. perform the Agent's task-wide image acquisition preflight below before capture;
+4. execute only the current route command.
 
 An explicit request to capture, ingest, fetch, or read named file/remote
 documents grants source-read permission for those named modules. A mention,
 possible-source discussion, or register-only request does not. An explicit
 refusal always wins.
 
-When that permission is already present in the conversation, execute the
+When that permission is already present and the Agent's image acquisition
+preflight is complete, execute the
 Gate's returned authority-carrying command. In managed mode it runs the
 deterministic capture batch until the next real blocker; in ordinary mode it
 reevaluates status with `context.source-read`. The authority remains in the
@@ -48,12 +50,26 @@ boundary is a documentation site rather than plain Markdown, use the
 Context-provided processor/configuration diagnostic; do not invent rendered
 text or scan outside the confirmed boundary.
 
-For Lark reads, Context prefers the authenticated user identity. If that
-identity is unavailable because its credential is missing, expired, or cannot
-be refreshed, Context may retry the same registered source with the bot
-identity. It does not switch identity after a permission or missing-scope
-response. Once selected, the same identity is used for the document body and
-all embedded resources. If `docs +fetch` lacks the required `--doc-format`
+For Lark reads, set `CONTEXT_LARK_IDENTITY=user|bot` in the environment of every
+Context invocation that may read Lark; the default is `user`. A host/Bot prompt
+setting alone is not an exported environment variable. User mode reads only
+as the current user. Bot mode reads as the configured application, then retries
+once as the current user on a credential, scope or access denial. Subsequent
+reads in that capture use the fallback identity, including pagination, synced
+references, media, Sheets, Base and whiteboards. Network, rate-limit and format
+errors do not switch identity. The diagnostic records any fallback.
+
+This setting does not grant source access or change the intended audience of
+the resulting knowledge. Do not use another person's credentials. For direct
+host `lark-cli` reads, explicitly pass the selected `--as bot` or `--as user`
+on every business read, including Wiki resolution/listing and asset reads.
+If Bot access fails for the reasons above, retry that read once as user; stop
+and report a pending source if user authorization is required. In a managed
+runtime, use its authorization flow rather than repeatedly running `auth login`.
+Local help and embedded skill discovery require no document authorization;
+do not request user OAuth just to read a CLI guide.
+
+If `docs +fetch` lacks the required `--doc-format`
 capability, follow the returned `lark-cli update` recovery and rerun the same
 Route command; do not replace the capture with a hand-written export.
 
@@ -63,9 +79,67 @@ are available while another capture remains pending. Keep unresolved sources in
 pending capture after the active production/review work, before delivery completes.
 This does not mark failed captures as complete or waive source-read authorization.
 
-For image-heavy tasks, reuse the image choice in the work-start report. When
-source descriptors already show more than 30 distinct images across the task,
-resolve that question before bulk media acquisition. If counts emerge only from
-capture, stop further image processing to confirm the choice during planning.
-Use explicit image/GIF reference-only policies for exclusions; never substitute
-temporary signed media URLs as permanent public image links.
+## Image acquisition before capture
+
+The workflow instruction setting `DOCUMENT_IMAGE_CAPTURE` accepts `ON`, `OFF`,
+or a positive integer document threshold; when absent it defaults to **10**.
+Read it from the current task instructions or workspace guidance; a current
+conversation instruction takes precedence. This is an Agent-interpreted setting,
+not a CLI flag, environment variable, or SDK field. Hosts need not add a setting.
+
+| Setting | Acquisition policy |
+| --- | --- |
+| `ON` | Include images without a document-count cutoff, subject to existing resource limits and explicit exclusions. |
+| `OFF` | Capture bodies and image references only, regardless of document count. |
+| positive integer `N` | Include images under the existing policy for at most `N` documents; above `N`, use references only. |
+| absent | Use `10`. |
+
+For an invalid value, report it and use `10` rather than silently enabling all
+images or blocking the task. An explicit current-conversation request to include
+all images selects `ON` for this task; an explicit request to skip images selects
+`OFF`. If neither is present, use the configured setting. Do not write a temporary
+conversation override back into permanent workspace guidance.
+
+Count distinct documents
+selected for this task, including Wiki descendants and in-scope linked documents,
+not all historical workspace sources, API pages, retries, or output articles.
+Deduplicate shortcuts and repeated links. Splitting the task into batches or
+resuming it does not reset the count. Exactly 10 documents is within the default.
+
+Before starting capture, the Agent separately checks the document count (or known
+lower bound), setting, and effective image policy. This is an Agent self-check,
+not a user confirmation gate: do not ask a question or wait for a reply. Above the
+numeric threshold, automatically capture bodies with image references only; at or
+below it, retain the existing image policy. `ON` and `OFF` apply regardless of count.
+A generic
+"capture everything", source-read permission, managed mode, or an old bundle setting
+does not. Briefly report the count and applied policy in the work-start update and
+retain the decision in task notes; do not repeat the update for every document.
+
+If the count is unknown, first perform only authorized directory/metadata discovery
+without image downloads. If body reads are needed to discover linked documents,
+use a body-and-reference-only discovery pass within existing read authorization.
+With a numeric setting, unknown totals do not authorize image acquisition; `ON`
+does not require counting to enable images, and `OFF` never enables them.
+Recalculate the task total before enabling images under a numeric setting.
+If later expansion crosses the numeric threshold, apply the
+reference-only policy before further acquisition and report the change without
+waiting for a reply; retain any explicit all-image override and existing evidence.
+
+For `OFF` or above-threshold tasks without an `ON` override, apply
+`resources: { images: "reference-only", gifs: "reference-only" }` to the selected
+Lark capture declarations before executing the Route. Preserve unrelated sources,
+resource settings and existing stricter exclusions; inspect generated/dynamic
+configuration rather than assuming registration applied the policy. Host prefetch
+and import paths must honor the same choice, not download images before import.
+Other resource types retain their existing policies. Do not edit snapshots or
+invent unsupported flags; when a capture adapter cannot exclude images, explain
+that limitation and resolve the acquisition method before downloading.
+
+Preserve source links/placeholders and report skipped images as not acquired or
+interpreted. This is a workflow instruction using existing resource policies, not
+a new CLI hard gate or automatic document counter. For tasks that still include
+images, retain the existing handling workflow for more than 30 distinct images
+and reuse an explicit choice; do not invoke it for images skipped by this
+preflight. Never substitute temporary signed media URLs as permanent
+public image links.

package/providers/context/resources/procedures/document-code-investigation.md

@@ -0,0 +1,45 @@
+---
+id: procedure.document-code-investigation
+kind: procedure
+mediaType: text/markdown
+---
+
+# Assess code needs during document investigation
+
+Register the requested document sources before preparing their article plan.
+Continue authorized document capture even when unrelated registered repositories
+are unavailable. Existing workspace coverage does not select the current task's
+code dependencies.
+
+After reading outlines, representative bodies and related approved articles,
+group the reader topics and assess whether code investigation is needed before
+submitting the plan and work-start report. Large collections can be investigated
+in batches while capture continues. If the expected production plan exceeds 30
+article tasks, explicitly perform this assessment before bulk writing. This is
+Agent guidance, not a CLI validation threshold. Wiki nodes, documents and article
+tasks are different counts; more than 30 document nodes calls for early batched
+investigation, not an assumption of 30 articles or automatic repository recovery.
+
+Code evidence is useful when the reader task requires implementation behavior,
+call chains, interface constraints or debugging entry points, or when incomplete
+or conflicting documents require checking code. Explicit file, symbol or API
+references strengthen that assessment. A repository link or service name alone
+does not establish a dependency. Keep document-supported topics independent.
+
+Use existing plan briefs or report prose to explain which reader questions need
+which registered modules and why; no separate assessment receipt, score or
+required schema field is needed. If code is needed, inspect its availability with
+`context source recovery-plan <registered-name> --format json`. Follow
+[repository source recovery](repository-source-recovery.md), reusing applicable
+authorization and local checkouts. Prefer the smallest sufficient registered
+module scope. Expand to a shared physical checkout or whole repository when
+cross-module calls, common dependencies or uncertain boundaries require it;
+do not substitute a different remote or pinned commit.
+
+If recovery is unavailable, continue independent document tasks. Keep questions
+requiring unread code pending, or narrow the article to supported document
+claims and explain its limits. Do not present inferred implementation as verified
+code, delete configured sources, or clear a stage to bypass unavailable evidence.
+After recovery, follow its returned preparation command and current task handles;
+retain accepted work instead of restarting the task. Reassess code needs when
+later reading exposes a dependency that the initial sample did not show.

package/providers/context/resources/procedures/repository-source-recovery.md

@@ -17,7 +17,9 @@ remote and pinned commit, so one physical checkout can restore every module in
 that group. Do not create one clone per module. Groups already marked `ready`
 need no decision; when every group is ready, the plan has no recovery action.
 
-For each missing physical checkout, ask the user to choose one option:
+For each missing physical checkout, reuse an applicable choice and authorization
+already supplied in the current task. If the recovery choice is unresolved, ask
+the user to choose one option:
 
 1. provide an existing local Git checkout;
 2. name a bounded directory that the Agent may scan, then choose one of the
@@ -43,3 +45,8 @@ Repository sources are ready only when every selected module resolves to the
 registered remote and pinned commit, every registered subpath exists, no local
 path was overwritten, Context materialization succeeds, and the selected recovery
 plan reports them ready.
+
+Recovery failure does not stop unrelated document capture or independent article
+tasks. Preserve pending code-dependent questions and follow the returned recovery
+diagnostic. After a successful restore, follow `next_action` to refresh affected
+production material; keep accepted articles and use the newly returned handles.

package/providers/context/resources/procedures/source-capture-detailed.md

@@ -71,6 +71,12 @@ settled scope, not a new confirmation step.
 
 ## Workflow
 
+Before any capture or host prefetch, follow the task-wide image threshold and
+Agent self-check in [Document capture](document-capture.md#image-acquisition-before-capture).
+Do not ask the user to confirm this default or infer an all-image override from
+source-read permission. Apply the effective resource policy before a managed batch;
+registration defaults must not silently re-enable excluded images.
+
 Capture is entirely CLI-driven — your role is to register the right source,
 declare the matching capture phase, run the current `context run capture:*`
 command, and preserve its machine fields while summarizing the outcome in the

package/providers/context/skills/work-production-stage/SKILL.md

@@ -19,6 +19,10 @@ where needed to decide reader topics, source grouping and writing batches.
 Indexer guidance discovers a skeleton, not a mandatory deep parse or a ledger
 of every symbol. Declare relevant available skills and record planned uses;
 do not collect skill hashes, versions or article ownership credentials.
+After reading representative documents and grouping reader topics, apply
+[document and code investigation guidance](../../resources/procedures/document-code-investigation.md)
+before submitting the article plan. Revisit it when new implementation questions
+appear during writing. This assessment adds no approval gate or required field.
 Submit the plan before presenting the work-start report. Planning does not
 authorize writing or approve the report.