docs: sync v0.5 status and ost pin

README.md

@@ -1,4 +1,8 @@
-# usd-http-resolver
+# OpenUSD Http Resolver
+
+[![CI](https://github.com/animu-sphere/usd-http-resolver/actions/workflows/core-ci.yml/badge.svg)](https://github.com/animu-sphere/usd-http-resolver/actions/workflows/core-ci.yml)
+[![License](https://img.shields.io/github/license/animu-sphere/usd-http-resolver)](LICENSE)
+[![OpenUSD](https://img.shields.io/badge/OpenUSD-26.08-4C78A8)](docs/compatibility/OPENUSD.md)
 
 An OpenUSD asset resolver for `http://` and `https://`, and the I/O substrate
 beneath it.
@@ -29,12 +33,14 @@ index, and the chunks actually in view — not 10 GB.
 
 ## Status
 
-**`v0.4.0` is released: a `UsdStage` opens over HTTP, a clustered read of a
+**`v0.5.0` is released: a `UsdStage` opens over HTTP, a clustered read of a
 remote asset costs three requests where it cost eighteen, and a process that
 starts cold pays nothing for a window an earlier one fetched. The read contract,
 the local backend, the shared boundary suite, the hostile-server corpus, the
 HTTP backend, the `ArResolver` bundle, the block cache, identity exposure, and
-the on-disk cache tier are in the tree and passing.**
+the on-disk cache tier are in the tree and passing. The aggregate resolver
+product's installed-artifact probe opens a remote root layer and its relative
+child, and requires a successful byte-range request.**
 
 That ordering is the point. `v0.1.0` shipped a local file reader, which is not
 interesting; what was interesting is that it arrived with the harness that makes
@@ -130,9 +136,10 @@ the per-format compatibility matrix, are in §4 of
 
 ## First consumer
 
-`usd-pointcloud-plugins` reads COPC over this resolver in `v0.5.0` — with no
-HTTP code of its own, no build dependency, and no change to its COPC reader. If
-it needs one, the abstraction leaked and the fix belongs here.
+`usd-pointcloud-plugins` is the first planned consumer integration in
+`v0.6.0`. It should read COPC over this resolver with no HTTP code of its own,
+no build dependency, and no change to its COPC reader. If it needs one, the
+abstraction leaked and the fix belongs here.
 
 ## Building
 

docs/architecture/WORKSPACE.md

@@ -357,14 +357,14 @@ member descriptor exists.
 11. A directory is created when its first tested capability exists.
 12. A release that changes I/O behavior records a metrics baseline.
 
-## Composed-runtime packaging (OST v0.22.8 dogfood)
+## Composed-runtime packaging (OST v0.22.10)
 
 The aggregate membership is explicitly pinned by `workspace.release_members` in
 `openstrata.toml`. Component-owned acceptance probes are installed under
 `share/usd-http-resolver/probes` through file-specific `workspace.install_data` mappings.
 They exercise the packaged payload and are not runtime dependencies. Probe output
 is caller-owned evidence and must not modify the immutable composed prefix.
 
-Package producers need the aggregate payload projection correction found during
-OST v0.22.8 dogfooding; stock ost 0.22.7 products advertise install sources absent
-from their archive inventory. CI source builds remain pinned to released 0.22.7.
+Package producers need the aggregate payload projection correction first found
+during OST v0.22.8 dogfooding. Current source builds are pinned to released
+OST v0.22.10.

docs/compatibility/OPENUSD.md

@@ -21,7 +21,7 @@ runtime:
 | --- | --- |
 | Target OpenUSD version | 26.08 |
 | Accepted range | `>=26.08,<27.0` |
-| OpenStrata CLI | 0.22.7 |
+| OpenStrata CLI | 0.22.10 |
 | OpenStrata platform / profile | `cy2026` / `usd` |
 | C++ standard | C++17 |
 | CMake | 3.23 or newer |

docs/guides/BUILDING.md

@@ -4,16 +4,17 @@ This repository is an OpenStrata project. It builds either through `ost`, which
 resolves and composes a certified OpenUSD runtime, or through plain CMake — and,
 for everything under `libs/`, with no OpenUSD at all.
 
-Status: `libs/usd-asset-io`, `libs/usd-asset-local`, and the shared boundary
-suite in `tests/` build and test today. No bundle exists yet, so the `ost
-plugin` steps below are the workflow the first bundle lands into rather than
-something you can run.
+Status: `libs/usd-asset-io`, `libs/usd-asset-local`, the shared boundary suite
+in `tests/`, and the `plugins/http-resolver` bundle build and test today. The
+v0.5.0 aggregate product and its packaged probe are covered by the release
+workflow; the `ost plugin` commands below cover bundle and workspace
+verification.
 
 ## Requirements
 
 | Tool | Version |
 | --- | --- |
-| `ost` | 0.22.7 or newer |
+| `ost` | 0.22.10 or newer |
 | CMake | 3.23 or newer |
 | Ninja | any recent |
 | A C++17 compiler | MSVC 143, Clang, or GCC |
@@ -29,7 +30,7 @@ ost doctor
 ost library build libs/usd-asset-io        # one descriptor-owned library
 ost library test  libs/usd-asset-io
 
-ost plugin build http-resolver             # once the bundle exists
+ost plugin build http-resolver
 ost plugin test  http-resolver
 ost plugin test --workspace --up-to 4
 ```

docs/roadmap/implementation-status.md

@@ -211,7 +211,7 @@ therefore ship unexercised.
 | Component-owned acceptance probe, installed with the product | Done — `share/usd-http-resolver/probes/packaged_probe.py`, mapped in as product data so acceptance runs from the artifact |
 | Probe proves registration *and* transport, not just loading | Done — a remote root layer, a relative child layer followed to a second remote layer, and a byte-range request that must succeed |
 | No producer build directory on any path the probe uses | Done — that separation is the reason the probe exists rather than reusing `httpResolver_stage` |
-| OpenStrata 0.22.8 release pins, generated CI refreshed | Done — canonical OpenUSD 26.08 pins regenerated on the v0.22.8 line |
+| OpenStrata 0.22.10 release pin, generated CI refreshed | Done — current CI and canonical OpenUSD 26.08 pins use the v0.22.10 line |
 | Reproducibility: two identical package operations, same SHA-256 inventory | Done — a release gate, walked; platform digests live in the generated release evidence rather than in the source record |
 | Behavior unchanged from `v0.4.0` | Asserted — transport, validator, cache, diagnostics, and configuration contracts all carried forward; the I/O baseline is `v0.4.0`'s, because the reader and cache algorithms did not move |
 | OpenStrata formation composition | Deferred to `v0.7.0` — a product exists; a formation pinning this resolver and a consumer bundle by digest against one certified runtime does not |

libs/usd-asset-cache/README.md

@@ -231,8 +231,10 @@ nothing.
 
 - `v0.4.0`: persistence, admitted per asset by validator strength, and the
   identity exposure that goes with it (CACHE.md §8).
-- `v0.5.0`: the first measurement over real distance. The block size chosen
+- `v0.5.0`: the packaged resolver product and artifact-owned acceptance probe;
+  the cache algorithms and their defaults do not change.
+- `v0.6.0`: the first measurement over real distance. The block size chosen
   here is chosen on request counts under a stated premise about round-trip cost
-  that loopback cannot price; `v0.5.0` is where the premise gets tested.
-- `v0.6.0`: cache configuration through `ArResolverContext`, so that two stages
+  that loopback cannot price; this is where the premise gets tested.
+- `v0.7.0`: cache configuration through `ArResolverContext`, so that two stages
   in one process can have two policies.

openstrata.ci.yaml

@@ -34,11 +34,11 @@ schema: 1
 bootstrap:
   ost:
     # Canonical OpenUSD 26.08 SDK leaves, produced with ost 0.22.3.
-    # ost 0.22.8 consumes their current manifest and supports locked composition.
+    # ost 0.22.10 consumes their current manifest and supports locked composition.
     # Windows/Linux use gl; macOS arm64 uses metal. Archive and OCI identities
     # are separate pins. This replaces the retired pre-canonical artifacts and
     # the 0.21.0 schema-compatibility workaround (historical OST report 03).
-    version: "0.22.8"
+    version: "0.22.10"
     repository: animu-sphere/open-strata
 
 runners:
@@ -148,6 +148,8 @@ cells:
       expected_oci_digest: sha256:8d2d599489ee623044f4360ad3a2b979e2966e6ad2f949e5aa9083570e59cc4f
     platform: cy2026
     profile: usd
+    require_openusd: cy2026/linux/x86_64/gl
+    require_openusd_version: "26.08"
     host_python: "3.13"
     host_packages:
       apt: [libx11-dev, libxt-dev, libxext-dev, libgl1-mesa-dev, libcurl4-openssl-dev]
@@ -163,6 +165,8 @@ cells:
       expected_oci_digest: sha256:8486d94286b60cc20b53c926757f1802442569f9d9a7eecdadf2c9895b2efeb6
     platform: cy2026
     profile: usd
+    require_openusd: cy2026/macos/arm64/metal
+    require_openusd_version: "26.08"
     host_python: "3.13"
     verify: test
 
@@ -201,6 +205,8 @@ cells:
     bundle: plugins/http-resolver
     platform: cy2026
     profile: usd
+    require_openusd: cy2026/linux/x86_64/gl
+    require_openusd_version: "26.08"
     host_packages:
       apt: [libx11-dev, libxt-dev, libxext-dev, libgl1-mesa-dev, libcurl4-openssl-dev]
     up_to: 1
@@ -215,4 +221,6 @@ cells:
     bundle: plugins/http-resolver
     platform: cy2026
     profile: usd
+    require_openusd: cy2026/macos/arm64/metal
+    require_openusd_version: "26.08"
     up_to: 1

tests/README.md

@@ -55,7 +55,7 @@ Nothing in it chooses a default from wall clock. Loopback has no round-trip time
 worth the name, and the whole argument for merging small reads is about a
 round-trip time this harness cannot produce; the defaults come from the request
 counts and the byte counts, under a premise about round-trip cost that the
-record states outright and `v0.5.0` is where it gets tested.
+record states outright and `v0.6.0` is where it gets tested.
 
 ## The baseline