v0.3.11

• Under the explicit Python runtime policy, L20 speed, current-limit and fault-clear writes end at physical dispatch without retaining Native optional reply decoders or read-key reservations. Subsequent reads do not wait for an unused reply window.
• Python-policy response matching starts at physical send claim, using Session event order when available and timestamps as a compatibility fallback. Replies received before a queued read is claimed cannot complete that read. Idle L20 write replies are not published as telemetry. Classic L20 replies have no transaction identifier: a same-opcode late write reply received after a read is claimed remains indistinguishable from a read response, as in the Python reference.
• Keep Native optional feedback, existing public C++ APIs, frame contents and inter-frame delays unchanged. No additional sender or scheduling queue is introduced.

The minimum and paired transport remain v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68). Rebuild the SDK and downstream consumers to use the fix; no public structures or signatures are added in this patch.

Validation is offline and covers missing and late replies, repeated write/read sequences, close/reopen, explicit recovery, queued-read claim boundaries and device isolation on a shared Session. No new physical-device acceptance is claimed. Deployed firmware reply timing, adapters/drivers, physical timestamps and multi-device sustained operation require separate hardware validation. O20, Python bindings and hand/arm integration are outside this patch.

Source: eb060911edeb20113fa17e88e3e06715b67033be. The signed v0.3.11 tag identifies this commit. PR #25 contains the implementation and regression coverage.

Validation:

• Eight targeted offline cases / 123 assertions; all ten local CTest groups passed in the final complete run (88.17 seconds). The queued-read regression fails without the claim filter, returning an old write value instead of the new read reply.
• 38 related tests passed against Python reference source 3a63e723b9a0046e9a60a58339b9f064ef5b70fd. Separate Python-runtime and installed-C++ probes confirmed dispatch-only write/read completion with different response-window overrides. Eighteen documented installed-consumer examples compiled. The probe timings describe the offline setup. Hardware throughput and latency require separate validation.
• Merged-source CI: Linux and Windows Debug/Release, ASan/UBSan, public headers, documented examples and consumers.
• Merged-source prebuilt consumers: x86_64 and aarch64 with transport v0.3.4.
• Complete merged-source analysis: A+, score 100, zero diagnostics, complete formatting and analyzer coverage.
• Explicit tag-based package build: twelve targets. All 26 assets were downloaded and independently checked for API SHA-256 and size, package source/version/dependency identity, object architecture and archive checksums before publication.

Validation record: The first PR Windows x64 Release attempt timed out in the unchanged Native L20 subcase of "HandDevice cache and validation stay available during a blocked send". An unchanged selected-job rerun passed, as did the independent Windows x64 Release package validation. Source behavior, assertions and timeout budgets were retained; the underlying cause was not established. Earlier local runs also encountered intermittent failures in unchanged Native fixtures; their follow-up checks and the final complete run passed. These records do not establish hardware timing behavior.

v0.3.10

• Treat missing verified angle-command feedback as a logical timeout under the Python policy. Preserve the command's physical Dispatched outcome, exclude it from successful transaction latency and retain its error count until a complete logical success. Native optional-feedback accounting is unchanged.
• Project Python-policy polling counters independently of native transaction totals. Inactive plans report zero polling counters; active reads, verified angle command feedback and disabled optional queries retain their distinct accounting.
• Measure complete logical transactions from their first transport admission through successful completion, including transport waiting and multi-stage execution. Partial stage success does not add a completed logical transaction or clear a prior logical error. Keep the latest 2048 successful durations across close, reopen and recovery, with Python nearest-even percentile selection. Native latency accounting remains unchanged.
• Add Python-policy force-field health rows with requested, shared-query and achieved rates, query identity and observed support. Optional PollingChannel::force_fields entries select named fields at millisecond intervals. Shared queries use the fastest requested interval; field rows must not be summed as physical traffic. L20Lite scalar categories remain independent. Native configurations reject the new selector.

Existing C++ method signatures and Native defaults remain available. Public structure additions require rebuilding the SDK and downstream consumers with matching headers and libraries; binary compatibility with earlier headers is not claimed. The minimum and paired transport remain v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68).

Validation covers offline lifecycle, polling, command feedback, multi-stage failure, field-query resolution, bounded latency windows, source/prebuilt/installed consumers, sanitizers, complete static analysis and the twelve-target release matrix. No new physical-device acceptance is claimed. Deployed adapters/drivers, firmware, physical timestamps, mixed CAN/FD topology and multi-device duration tests require separate hardware validation. O20, Python bindings and hand/arm integration are outside this patch. Tactile-stage angle scheduling, extended identity parity and optional automatic reconnect remain separate work.

v0.3.9 was tagged but not published. Post-merge verification found that missing command feedback could add a successful latency sample; this patch includes that correction. The v0.3.9 tag remains unchanged.

Source: 5aa263c. Paired transport: 0e190032d65ed0b315337b45b3779cf4f650ed68 (v0.3.4).

Offline evidence: 18 added health cases / 177 assertions, all 10 local CTest groups, 59 reference Python runtime tests, and 6144 rolling-window updates compared with an independent Python oracle and the reference RuntimeMetrics. These checks do not establish physical-device behavior.

Validation records:

• Merged-source CI
• Released-transport consumers
• Complete A+ analysis
• Native package acceptance and draft verification

The published set contains twelve native packages, twelve per-package checksum files, release-manifest.json and SHA256SUMS.txt (26 assets). Downloaded draft bytes were independently checked against API SHA-256 digests and sizes, source identities, archive checksums and object architectures before publication.

An initial merged-source Windows x64 Release run timed out in the existing Native L20 current-limit fixture. The selected job passed when rerun on the unchanged source, assertions and timeout settings. The initial failure remains in the workflow history; its root cause was not established.

v0.3.8

• Remove unclaimed Python-policy poll demand, release poll observers and wait up to shutdown_timeout for already-claimed polls. Preserve coalesced active readers and other devices sharing the Session. A stop timeout leaves the device open with its polling intent disabled, including after recovery.
• Project Python-policy health and channel-age generations onto the facade's successful open/recovery generation, matching retained snapshots across close, reopen and recovery. Physical Session generations remain separate. Native health projection is unchanged.
• Record an optional query's final UnsupportedFirmwareError in health without counting its underlying failure twice. Count post-decode unsupported results once when native metadata reads succeeded. Associate the update with that native failure's revision so a later error cannot be overwritten. Successful initialization clears prior errors only when no newer native error followed its last successful transaction.
• Add optional route_queue_capacity to device and discovery runtime configuration for the Python policy. The unset Python value selects 256 routed receive frames per device; a positive override reaches the existing Runtime inbox and survives recovery. Native defaults remain 512, and Native configurations reject the new Python-only override. Invalid capacity is rejected before transport creation. No receive layer or physical send queue is added.

Existing C++ method signatures and Native defaults remain available. Public configuration additions require rebuilding the SDK and downstream consumers with matching headers and libraries; binary compatibility with earlier headers is not claimed. The minimum and paired transport remain v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68).

Validation uses offline FakeTransport lifecycle, coalesced-reader, shared-Session isolation, error-ordering and inbox-overflow regressions, source/prebuilt/installed consumers, sanitizers, complete static analysis and the twelve-target release matrix. These checks do not add physical-device acceptance. Deployed adapters/drivers, firmware, physical timestamps, mixed CAN/FD topology and multi-device duration tests require separate hardware validation. O20, Python bindings and hand/arm integration are outside this patch; tactile-stage angle scheduling, extended identity parity and optional automatic reconnect remain separate work.

The no-sensor O30 regression exposed a signed overflow while reconstructing an unbounded Python-policy composite deadline. That response-window path now uses the saturating deadline helper; Native deadline calculation remains unchanged.

Validation for hand source 65477aa8393e145b227063ad421ca8d395b4153e and paired transport source 0e190032d65ed0b315337b45b3779cf4f650ed68:

• CI: GCC Debug/Release, Windows x64 Debug/Release and ASan/UBSan passed. Released-transport consumers passed on x86_64 and ARM64.
• Complete static analysis: A+, score 100, zero diagnostics. The downloaded report was independently checked: 613/613 formatting files and 358/358 units for each of cppcheck and clang-tidy.
• Release acceptance: all twelve targets, asset collection and exact-source draft verification passed. The packages cover Ubuntu 20.04/22.04/24.04/26.04 x86_64/ARM64 Release and Windows x64/ARM64 Debug/Release.
• Local regression coverage includes 16 new cases with 143 assertions, all eight CTest groups, the installed consumer and compilation of 18 documented examples. Reference checks include 30 executable observations, 31 same-input outgoing-frame vectors and 40 Python runtime tests using Python SDK commit 3a63e723b9a0046e9a60a58339b9f064ef5b70fd. This reference pin does not identify the deployed build or establish complete Python behavioral equivalence.

All 26 draft assets were downloaded and independently verified against the twelve target identities, source revisions, object architectures, archive checksums and GitHub asset digests. Use release-manifest.json and SHA256SUMS.txt to verify the selected package.

Validation history: an intermediate PR revision exposed the Python-policy O30 deadline overflow described above. Another intermediate revision failed Windows ARM64 communication/discovery acceptance. After Python helper ownership was separated from Native calls, both configurations passed on the final PR and release sources. An unchanged v0.3.7 comparison also passed all twelve targets. These results do not establish the underlying cause of the intermediate ARM64 failures or a compiler defect. The adjacent-Session-parameter analyzer finding was corrected without disabling its diagnostic.

See PR #22 for implementation and regression coverage. Select the matching platform, architecture and Windows runtime configuration; vendor drivers are not bundled. Existing physical-device acceptance records retain their original scope.

v0.3.7

• Extend the explicit RuntimeBehavior::kPython policy with publication-based channel sequences and coherent retained snapshots. Cache readers do not advance acquisition counters. Identity reads carry independent sample metadata; closed Python-policy caches retain identity evidence with invalid metadata.
• Add wait_for_cached_snapshot(previous, timeout, wake_on_command_feedback) without overloading the existing two-argument wait signature. Filtered waits use the last non-command publication, retain command feedback in the current cache, and wake when their device generation ends. These waits perform no device I/O.
• Reject matching malformed Classic tactile rows in the Python policy, including invalid lengths, row markers and conflicting duplicates. Unrelated routes and opcodes remain unmatched. Correct force aggregation so one query replaces its own fields and scalar raw data follows the latest scalar query while matrix evidence from other queries remains available.
• Read L20Lite fingertip scalar categories independently in the Python policy. Track firmware-optional failures per channel and generation: only sent, unverified optional queries can degrade after repeated response/protocol failures; a previously successful channel is not disabled by later response loss. Disable only that channel and restore probing after explicit recovery.
• Add query-only, command-feedback-only and adaptive angle feedback policies to unified polling. Only verified current-state command feedback fulfills the requested angle rate. Equivalent plans are no-ops. Polling uses the existing native runtime and transport Session; no second physical send queue is added.

Native behavior remains the default. The policy covers Classic L6, O6, L20Lite, L20, L30 and explicitly selected O30 V25. Existing C++ method signatures remain available; public metadata and configuration additions require rebuilding the SDK and downstream consumers with matching headers and libraries. The minimum and paired transport remain v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68). Binary compatibility with earlier headers is not claimed.

Validation uses offline FakeTransport regressions, Python reference contracts, native/source/prebuilt/installed consumers, sanitizers, complete static analysis and the twelve-target release matrix. No new physical-device acceptance is claimed. Adapter/driver/firmware, physical timestamps, deployed tactile layouts, mixed CAN/FD channels and long-duration multi-device behavior require separate hardware validation. O20, Python bindings, automatic reconnect and hand/arm integration are outside this patch. Extended identity raw/version field parity and O30 field-by-field identity requests remain separate work.

Build and verification

Source commit: 1a8063be54f80bf9b9d5421cf5b2cca690b0d5a0 (PR #21). The signed v0.3.7 tag identifies this commit.

• Merged-commit CI: GCC 13 and Windows x64 Debug/Release, ASan and UBSan passed.
• Released-transport consumers: x86_64 and aarch64 passed against transport v0.3.4.
• Complete static report: A+, score 100, zero diagnostics; 612 format files and 357 compilation units for each analysis tool. The downloaded report passed the completeness validator.
• Release acceptance: all twelve package targets, collection and draft-release checks passed. The 26 downloaded draft assets were independently checked for package/source identities, object architectures and checksums before publication.

Release validation initially reported an intermittent failure in the unchanged Native L20Lite stop/restart polling test. The same assertion reproduced with the unmodified v0.3.6 SDK and original test in concurrent focused runs (2/500 with eight workers on the local host); 500 serial runs passed. The isolated package retry passed without source, test, configuration or assertion changes. Native cancellation and late-response guards are unchanged.

Local offline verification also passed: 8/8 CTest groups; 74 phase-two/phase-three tests with 605 assertions; installed-consumer checks; compilation of 18 documented examples; 30 cross-language observations; 31 outgoing-wire vectors; and 80 Python regressions against reference commit 3a63e723b9a0046e9a60a58339b9f064ef5b70fd. This reference pin does not identify the deployed Python build. These checks do not add physical-device acceptance.

v0.3.6

• Add an explicit RuntimeBehavior::kPython policy for the unified facade. The native policy remains the default. Classic L6, O6, L20Lite, L20, L30 and explicit O30 V25 are in scope; L6 HOP and mounted O30 retain their existing native path.
• Coalesce equivalent active reads, identity refreshes and polling by device. Callers retain independent wait deadlines and bounded admission. A result or error published after a caller deadline does not replace that caller's timeout. Unclaimed reads without remaining demand are discarded; polling demand remains independent of internal waiter timeouts. A timed-out reader cannot cancel another reader's transaction. Latest setters remain replaceable until the device transaction lane is available. The existing transport Session remains the physical sender.
• Start required response windows after physical batch completion, retain a separate recovery execution deadline, and buffer early replies/errors until delivery is known. Classic identity uses one four-request batch with the Python 5 ms spacing. Protocol defaults and an explicit ordinary response-window override are available without changing native timeout defaults. Identity windows use identity_timeout independently; legacy Manager waits do not shorten the selected policy.
• Reject matching malformed Classic vector, bank and identity responses in the opt-in path. Preserve the last successful cache and apply finite late-response guards. Unrelated routes and opcodes remain unmatched.
• Under the opt-in policy, serialize endpoint admission and receipt retention with close. Drain all frames of an already claimed physical batch, including read and required-ACK batches; a required-ACK command still fails closed. Preserve the physical outcome of a claimed dispatch-only command during close. Unclaimed commands fail closed; a required-ACK command cannot become successful from delivery alone. No command is replayed on reopen or recovery. Closed runtime references cannot gate or cancel a reopened profile endpoint.
• Let concurrent callers join an explicit recovery cycle, with independent waits. Retry response-window loss only, after an endpoint quiet period, without reopening the transport for every initialization attempt. Protocol, identity and transport failures stop the cycle. Failed preparation cannot start a later reconnect; explicit recovery is also allowed from Faulted.

The minimum and paired transport remain v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68). Rebuild the SDK and downstream consumers with matching headers and static libraries. Existing signatures and native default behavior remain available; binary compatibility with earlier headers is not claimed.

Validation is offline unless a separate hardware record states otherwise. FakeTransport regressions and pinned Python behavior vectors cover the changed transaction and lifecycle paths. Native/source/prebuilt/installed consumers, sanitizers, static analysis and the twelve-target package matrix remain release gates. No new physical-device acceptance is claimed. Deployed adapters, drivers, firmware, timestamps and mixed-device channel behavior still require hardware acceptance. O20, Python bindings, automatic reconnect/adaptive feedback parity and hand/arm integration are not included.

Source: hand commit d7701bad9d01f9ff72c2e78d1c0c5bc5c39df0da, signed tag v0.3.6. Python behavior reference: 3a63e723b9a0046e9a60a58339b9f064ef5b70fd; this identifies the comparison source, not a deployed hardware acceptance record.

The changed paths have 46 focused regression cases (334 assertions), alongside the full offline suites, 30 executable native/profile/Python observations and 31 same-input wire vectors.

Validation records: merged-source CI, prebuilt transport consumers, complete static report, tagged twelve-target package and draft verification.

Packages cover Ubuntu 20.04/22.04/24.04/26.04 x86_64 and ARM64 Release, plus Windows x64 and ARM64 Debug/Release. Select the matching platform and runtime, and verify the manifest and SHA-256 checksums. Vendor drivers are not bundled.

v0.3.5

• Add unified polling for measured speeds on L20Lite, L20, L30 and O30, and voltages on O30, using the existing native polling and Session send path.
• Correct explicit O30 V25 torque-limit and motion-time freshness to 100 ms and 300 ms. Apply these bounds to projected snapshots and health. Mounted V36 remains unchanged.
• Add refresh_device_info() for an identity read that preserves queued commands, polling and the current generation on success. Timeout/protocol failures leave readiness unchanged; identity mismatch or transport loss faults admission and invalidates the readable cache. Existing get_device_info() and verify_identity() retain their earlier lifecycle behavior.
• Add get_cached_snapshot() for zero-I/O reads in all lifecycle states. Initial supported vectors are shaped and invalid. Generation loss retains raw evidence with invalid metadata, clears joint projections and omits identity. Physical connection loss or a changed shared Session generation cannot revalidate old evidence. Successful open/recovery replaces the cache; failed initialization cannot publish a partial baseline. Existing get_snapshot() still requires Ready.
• Keep hand and transport patch versions independent within the supported minor line. Source, prebuilt, installed-consumer and release checks retain the explicit transport 0.3.4 minimum.

The paired transport remains v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68). Rebuild the SDK and downstream consumers with matching headers and static libraries. Existing C++ signatures and native defaults remain available; no new binary compatibility guarantee is made.

Validation covers native builds, FakeTransport lifecycle/concurrency regressions, source/prebuilt/installed consumers, sanitizers, static analysis and the twelve-target package matrix. No new physical-device acceptance is claimed. The new polling channels, identity refresh, recovery/cache transitions and deployed adapter/driver/firmware combinations require separate hardware validation. Existing hardware records retain their original scope. O20, Python bindings, hand/arm integration and complete Python runtime equivalence are outside this patch.

Hand source: a20ac60eced4ad6f59e59cf779c970ecb870371f. Transport source: 0e190032d65ed0b315337b45b3779cf4f650ed68.

The release contains eight native Ubuntu 20.04/22.04/24.04/26.04 x86_64/ARM64 Release packages and four Windows x64/ARM64 Debug/Release packages. Select the matching platform and runtime. Vendor drivers are not bundled. See release-manifest.json and SHA256SUMS.txt for build identities and checksums.

Exact-source validation: CI, A+ quality gate, and twelve-package release validation.

v0.3.4

• Explicitly mark Classic L6/O6/L20Lite/L20 protocol frames as classic CAN so an FD-capable shared Session cannot promote them to FD through backend defaults. Existing classic-link behavior is retained.
• Add unified L6 position Kp/Kd, torque Kd, torque-output and limit-compensation commands and samples. Keep returned configuration bytes rather than substituting requested values. Torque output uses FIFO and bounded mode-transition guards derived from the Python reference.
• Add submit_operation() for nonblocking operation admission through the existing bounded FIFO; retain the earlier synchronous operation methods.
• Add asynchronous save_parameters_confirmed() for Classic L6/O6. Require the explicit CF 01 reply. Keep the existing save_parameters() behavior unchanged.
• Add L30 calibration as an ACK-gated disable, unlock and calibrate sequence. A failed stage prevents later writes; calibration does not re-enable the hand.
• Add an explicit O30 Python V25 layout for compact whole-hand commands and telemetry, including torque limits and motion time. Keep the default mounted V36 path and native Managers unchanged. Require complete response coverage and distinguish configuration queries from write ACKs.
• Add an explicit empty-polling-plan disable option, propagated through discovery. Keep existing no-argument polling defaults.
• Add same-input outgoing-frame vectors from Python SDK commit 3a63e723b9a0046e9a60a58339b9f064ef5b70fd, FakeTransport failure-boundary regressions and installed-consumer coverage. Record remaining migration requirements in the Python behavior baseline guide. O20 and Python hand/arm integration are not included.

The minimum transport dependency and bundled baseline remain v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68). Rebuild the SDK and every downstream consumer with matching headers and libraries. New public struct fields and variant alternatives do not preserve binary compatibility with older builds. Existing C++ method signatures, enum values and default protocol paths remain available.

Verification uses native builds, offline regressions, package consumers, sanitizers and static analysis. This release does not add physical-device acceptance. Persistence, calibration, torque-mode transitions, V25 behavior, deployment adapter/driver/firmware combinations and long-duration multi-device operation remain subject to hardware validation. Existing hardware records retain their original scope. The added interfaces do not establish complete Python behavioral equivalence or Python binding safety.

Hand source: 36010e7fb3eecd06cbbb0db3c067666bd98854a7. Transport source: 0e190032d65ed0b315337b45b3779cf4f650ed68.

The release contains eight native Ubuntu 20.04/22.04/24.04/26.04 x86_64/ARM64 Release packages and four Windows x64/ARM64 Debug/Release packages. Select the matching platform and runtime. Vendor drivers are not bundled. See release-manifest.json and SHA256SUMS.txt for build identities and checksums.

Exact-source validation: CI, A+ quality gate, and twelve-package release validation.

v0.3.3

• Require transport 0.3.4 or newer in the 0.3 minor line for source builds, parent-provided targets, prebuilt selection and installed CMake consumers. Keep the version floor in one CMake definition, independent of the hand patch version. Existing same-minor compatibility and the requirement not to use a transport patch older than hand are retained.
• Reject an older transport target already loaded by the parent of an installed consumer, even when a compatible package file is found.
• Apply the same floor to release validation and package collection. Add source/prebuilt/parent-target rejection tests and installed-package minimum, older-patch and newer-patch regressions using the exported package template.
• Retain existing C++ API signatures, runtime defaults, protocol behavior and scheduling. The supported legacy transport package and target aliases remain available.

Rebuild hand and downstream applications with matching transport headers and libraries. The release bundles transport v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68); do not mix the new headers with earlier objects or libraries. Binary ABI compatibility with earlier transport SDKs is not promised.

Native SDK builds, offline regressions, installed consumers, sanitizers and static analysis do not establish new physical-device acceptance. No new hardware/firmware/driver combination, physical timing or device soak is qualified. Vendor drivers are not bundled. Existing hardware records retain their original scope.

Packages and verification

The release contains eight native Ubuntu 20.04/22.04/24.04/26.04 x86_64/ARM64 Release SDKs and four Windows x64/ARM64 Debug/Release SDKs. Select the matching distribution, architecture and Windows CRT. Verify the signed v0.3.3 tag, release-manifest.json, SHA256SUMS.txt and the selected archive checksum. The manifest records both source revisions. Existing release assets are unchanged.

Hand source: f88198c770190b5e29aa2d2f8b443f027fee54ff; bundled transport v0.3.4 source: 0e190032d65ed0b315337b45b3779cf4f650ed68.

Validation

• Merged-source CI
• Published transport consumers
• Complete A+ report
• Twelve-package release validation and uploaded-byte verification

All runs identify the hand source above. The full report is A+ (100/100), with 588/588 formatting inputs and 345/345 translation units checked by each analyzer, without diagnostics. These results do not qualify host scheduling latency, physical CAN behavior or device performance.

v0.3.2

• Run the compiler-configuration compatibility matrix as a separate required CTest item, with its own process budget; retain the fast package audit and device-test limits.
• Read the current SDK version in package-compatibility validation instead of assuming 0.3.0.
• Use transport v0.3.4 (0e190032d65ed0b315337b45b3779cf4f650ed68) as the source and prebuilt validation baseline. The bundled SDK pairs these exact sources; standalone consumers retain the existing same-minor, minimum-version package policy.
• Give the interstage deadline-selection regression sufficient host-scheduling headroom while retaining its delayed reply beyond the angle budget. Production deadlines are unchanged.
• Update build, installation and release documentation. Device protocols, public hand API signatures, runtime defaults and scheduling are unchanged by this dependency update.

Rebuild the hand library and downstream applications with matching transport headers and libraries. Transport v0.3.4 adds public structure fields; do not mix old objects or libraries with the new headers. Binary ABI compatibility with earlier transport SDKs is not promised.

The release matrix builds and consumes native Ubuntu 20.04/22.04/24.04/26.04 x86_64/ARM64 Release packages and Windows x64/ARM64 Debug/Release packages. Verify the release manifest and checksums and select the matching platform and runtime. Vendor drivers are not bundled.

Offline tests, installed consumers, sanitizers and static analysis do not establish new physical-device acceptance. No new hand/arm hardware combination, driver/firmware combination, physical timing or long-duration device run is qualified by this release. Existing hardware records retain their original scope.

The v0.3.1 tag was created during release validation, which subsequently found a stale package-test version. No v0.3.1 Release was published; its tag is retained and this release uses a new immutable tag.

The SDK archives bundle hand and transport v0.3.4. Verify the signed v0.3.2 tag, release-manifest.json, SHA256SUMS.txt and individual archive checksums. Existing published assets are unchanged.

Validation

Hand source: a2f2219221e954d7774f7928910fcad6e22cd10c; transport v0.3.4 source: 0e190032d65ed0b315337b45b3779cf4f650ed68.

• Merged-source CI
• Published transport consumers
• Complete A+ report
• Twelve-package release validation and uploaded-byte verification

The exact-source report is full mode, A+ (100/100), with 588/588 formatting inputs and 345/345 translation units checked by each analyzer, without diagnostics. These checks do not qualify real-time host scheduling or physical-device performance.