test(cli): shared-files journey + .gitignore + docs (§11.6 files C5)

- collections.rs: shared_files_collection_is_read_write_across_the_subtree
  — a group declares kv+docs+files in one string-or-table manifest; authed
  app A creates a blob via files::shared_collection("assets").create(...),
  app B lists + gets the SAME bytes back across the subtree, a
  sibling-subtree app gets CollectionNotShared, collections ls shows all
  three kinds. Live-verified the bytes land under
  files/groups/<group_id>/assets/<id[0:2]>/<id>.
- .gitignore: ignore /crates/picloud-cli/data (the CLI journey harness
  spawns the server from that dir; the files journey is the first CLI test
  to write blobs).
- docs: design §11.6 (KV+DOCS+FILES shipped; topics/queue still deferred),
  sdk-shape (files::shared_collection), CLAUDE.md current-focus.

Full journey suite 117/117; schema blessed (55 migrations); workspace
clippy -D clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

docs/design/groups-and-project-tool.md

@@ -1044,31 +1044,38 @@ Resolved items now live inline next to their topic. What genuinely remains:
    real shared-scope authz model. Optionally, trigger/route **templates** (§4.5) if cardinality
    demands.
 
-   > **Shipped — §11.6 KV + DOCS slices (full shared read/write).** A group declares a collection
-   > group-shared (`[group]` manifest `collections = [...]`, owner-polymorphic marker table
+   > **Shipped — §11.6 KV + DOCS + FILES slices (full shared read/write).** A group declares a
+   > collection group-shared (`[group]` manifest `collections = [...]`, owner-polymorphic marker table
    > `0052_group_collections` with a `kind` discriminator); the data lives in a per-kind store keyed by
    > the owning `group_id` (NOT app — a shared row belongs to the group): `0053_group_kv_entries`
-   > (`kind='kv'`) and `0054_group_docs` (`kind='docs'`, the queryable-JSON store). Scripts read/write
-   > via the **explicit** `kv::shared_collection("catalog")` / `docs::shared_collection("articles")`
-   > handles (distinct `GroupKvHandle`/`GroupDocsHandle`; `shared` alone is a Rhai reserved word, hence
-   > `shared_collection`). The service resolves the owning group from `cx.app_id`'s ancestor chain
-   > **filtered by kind**, nearest-group-wins — **that walk is the isolation boundary**: a foreign
-   > app's chain never contains the owning group, so the name returns `CollectionNotShared`; a `kv` and
-   > a `docs` collection of the same name are distinct stores. **Trust model:** reads are open to any
-   > subtree script (anonymous public HTTP included — the declaration *is* the grant); **writes require
-   > an authenticated editor+** on the owning group (`GroupKvWrite`/`GroupDocsWrite`, fail closed on an
-   > anonymous principal). The docs slice reuses the `docs_filter` DSL — `build_find_query` was
-   > generalized on its owner column (`docs`/`app_id` vs `group_docs`/`group_id`), both compile-time
-   > literals, so the find SQL has one source. Declarative authoring uses the **string-or-table** form:
-   > `collections = ["catalog", { name = "articles", kind = "docs" }]` (bare string = `kv`). CASCADE on
-   > group delete; an app delete leaves the shared data. Live- + journey-validated for both kinds (app
-   > A writes, app B reads/finds; a sibling-subtree app gets `CollectionNotShared`).
+   > (`kind='kv'`), `0054_group_docs` (`kind='docs'`, the queryable-JSON store), and `0055_group_files`
+   > (`kind='files'`, the blob store — metadata in Postgres, bytes on disk under
+   > `<root>/files/groups/<group_id>/...`, a `groups/` infix disjoint from the per-app `files/<app_id>/`
+   > subtree so the existing recursive orphan sweeper covers both). Scripts read/write via the
+   > **explicit** `kv::shared_collection("catalog")` / `docs::shared_collection("articles")` /
+   > `files::shared_collection("assets")` handles (distinct `GroupKvHandle`/`GroupDocsHandle`/
+   > `GroupFilesHandle`; `shared` alone is a Rhai reserved word, hence `shared_collection`). The service
+   > resolves the owning group from `cx.app_id`'s ancestor chain **filtered by kind**, nearest-group-wins
+   > — **that walk is the isolation boundary**: a foreign app's chain never contains the owning group, so
+   > the name returns `CollectionNotShared`; a `kv`, a `docs`, and a `files` collection of the same name
+   > are distinct stores. **Trust model:** reads are open to any subtree script (anonymous public HTTP
+   > included — the declaration *is* the grant); **writes require an authenticated editor+** on the
+   > owning group (`GroupKvWrite`/`GroupDocsWrite`/`GroupFilesWrite`, fail closed on an anonymous
+   > principal). The docs slice reuses the `docs_filter` DSL — `build_find_query` was generalized on its
+   > owner column (`docs`/`app_id` vs `group_docs`/`group_id`), both compile-time literals, so the find
+   > SQL has one source; the files slice likewise generalized the atomic-write + checksum-on-read path
+   > helpers on an owner-relative dir, so the security-sensitive disk mechanics have one source.
+   > Declarative authoring uses the **string-or-table** form: `collections = ["catalog", { name =
+   > "articles", kind = "docs" }, { name = "assets", kind = "files" }]` (bare string = `kv`). CASCADE on
+   > group delete; an app delete leaves the shared data. Live- + journey-validated for all three kinds
+   > (app A writes, app B reads/finds/gets the same bytes; a sibling-subtree app gets
+   > `CollectionNotShared`).
    >
    > **Deferred (documented gaps):** write-triggers/events on shared collections (the "group trigger
-   > has no app to watch" problem); **files/topics/queue** shared collections (the `kind` column
-   > generalizes — files needs group-keyed disk paths); per-group total-size quotas + write-rate
-   > limits; CAS/`set_if`; app-declared collections. Multi-node tree-apply leans on the runtime
-   > backstop for no-op edges, as elsewhere.
+   > has no app to watch" problem); **topics/queue** shared collections (trigger-centric, same gap);
+   > per-group total-size quotas + write-rate limits; CAS/`set_if`; an operator admin API for shared
+   > blobs (scripts use the SDK; `pic collections ls` shows the marker — matches KV/docs); app-declared
+   > collections. Multi-node tree-apply leans on the runtime backstop for no-op edges, as elsewhere.
 
 ### 11.1 Re-sequencing review (post-Phase-3)
 

docs/sdk-shape.md

@@ -99,14 +99,18 @@ script. **Reads** are open to any subtree script (including anonymous
 public endpoints — declaring the collection *is* the grant); **writes**
 require an authenticated principal with editor+ on the owning group.
 
-**Docs too.** The same handle exists for the queryable-JSON store:
-`docs::shared_collection("articles")` returns a shared-docs handle with the
-full `create`/`get`/`find`/`find_one`/`update`/`delete`/`list` surface (same
-trust model + isolation boundary as the KV one). A `[group]` declares a
+**Docs and files too.** The same handle exists for the queryable-JSON store
+and the blob store: `docs::shared_collection("articles")` returns a shared-docs
+handle with the full `create`/`get`/`find`/`find_one`/`update`/`delete`/`list`
+surface, and `files::shared_collection("assets")` returns a shared-files handle
+with `create`/`head`/`get`/`update`/`delete`/`list` (Blob in/out) — both with
+the same trust model + isolation boundary as the KV one. A `[group]` declares a
 collection's store with a kind: `collections = ["catalog", { name =
-"articles", kind = "docs" }]` — a bare string is `kv`, a `{ name, kind }`
-table sets it. A `kv` and a `docs` collection of the same name are distinct
-stores. (Files/topics/queue shared collections are not yet implemented.)
+"articles", kind = "docs" }, { name = "assets", kind = "files" }]` — a bare
+string is `kv`, a `{ name, kind }` table sets it (`kv`/`docs`/`files`). A `kv`,
+a `docs`, and a `files` collection of the same name are distinct stores; group
+blobs live on disk under `<root>/files/groups/<group_id>/...`. (Topics/queue
+shared collections are not yet implemented.)
 
 ## Error convention