Final workspace validation

Validation scope: current coordinated working tree, Darwin arm64 host, release profile and locked dependencies. This is a source-candidate check, not a registry publication or a live network deployment. No wallet/node mutation is involved.

Current final rerun

After the mining, Neptune adapter, local-testnet1 and scoped formatting changes, RAYON_NUM_THREADS=4 RUST_TEST_THREADS=4 cargo test --workspace --all-features --release --locked passed 428 tests, with 0 failures and 3 ignored, across 55 test/doc-test binaries. Exit0, zero Rust warnings. Receipt: /tmp/trisha-workspace-final-current-rerun.log. This supersedes the earlier 409-test receipt below.

The first attempt exposed a test-only JSON Value to TOML serialization issue under serde_json arbitrary_precision. The owner replaced the fixture with native TOML values; typed production ProofFile serialization was unaffected. The full rerun includes both fixed Neptune policy proof workflows and the FIFO input regression. Initial failure: /tmp/trisha-workspace-final-current.log; narrow fix receipt: /tmp/trisha-policy-typed-toml-fix.log.

The three intentionally ignored tests are:

• genuine_transaction_prepare_and_mock_gateway_process (CLI deployment).
• genuine_custom_lock_transaction_and_binding_mutations (Neptune adapter).
• outer_proof_binds_compiled_recursive_sdk_to_public_claim (recursive SDK).

Their separately scheduled proof evidence is not inferred from this run. All three were subsequently invoked and passed: the recursive SDK outer proof and both genuine Neptune deployment tests. The actual isolated Neptune node then accepted the same canonical kernel into its mempool and rejected a changed kernel with the original proof. These separate receipts do not change the ordinary suite's 428 passed / 3 ignored accounting.

The separate default-feature CLI package run, RAYON_NUM_THREADS=4 RUST_TEST_THREADS=4 cargo test -p trisha --release --locked, also passed: 33 passed, 0 failed, 1 ignored, across7 binaries, exit0 and zero Rust warnings (/tmp/trisha-default-cli-final-current.log). This includes13 CLI unit tests, output-construction1, deployment1, wallet8, node-status1, recursive-input3 and source-pipeline6. The ignored test is the genuine transaction prepare/mock-gateway proof gate listed above.

Scoped formatting changed only 13 already changed/untracked Trident Rust files and 17 Trisha files, including the separate oracle workspace; those files now pass rustfmt. Unchanged pre-existing formatting differences in other files and sibling dependencies were not modified. git diff --check passed in Trident, Trisha and Joy; Joy scoped cargo fmt check passed. A whole-workspace formatting pass is not claimed. The post-format Trident workspace neural suite also passed 851 tests, zero failures/ignored/warnings, exit0: /tmp/trident-postformat-workspace-neural.log.

Earlier test commands

• cargo test --workspace --release --all-features --locked (/tmp/trisha-final-workspace-validation.log): exit101. Two failures in trisha-rs --test triton_compile: test_coin_compiles:434 and test_card_compiles:479 require a merkle_step opcode. The selected PLUMBv2 examples now use plumb.update_leaf with explicit pair/hash operations over the same20 siblings and assert_digest of the old root. Reported to the owner; the owner then authorized test-only corrections. The checks now execute an independent PLUMBv2 transition and reject changes to every coordinate of both roots, changed shared-path siblings, and path truncation; they also require update_leaf linkage and digest assertions. Production code is unchanged. Full rerun: /tmp/trisha-final-workspace-validation-rerun.log passed 409 tests, with 1 ignored, across 49 test/doc-test binaries. Exit0; zero Rust warnings. No Rust compiler warnings were emitted.
• Default CLI source/process tests: 19 passed, zero warnings, using cargo test -p trisha --release --locked --test source_pipeline --test neptune_wallet --test node_status --test deployment --test recursive_input (/tmp/trisha-final-default-cli.log). Counts: source6, wallet8, node-status1, deployment1, recursive-input3. This includes real small source proof/verify plus hostile-claim rejection, outside-checkout resources, and mock wallet IO.
• Ignored recursive proof tests are intentionally excluded from this run; their separate completed proof evidence belongs to the recursive backend audit.

Read-only capability and documentation review

Confirmed directly against code and owned target manifests:

• Trident's default target is nox; Trisha's own source commands select Triton by default. These are separate defaults and should remain explicit.
• Trisha target packages use schema1/compiler API2, Triton VM7.0.0 and native proof/claim version5. The proof envelope is stark-triton-v7; verification requires the current native version and default upstream security parameters.
• Joy packages likewise use schema1/compiler API2. Private execution uses joy-nox-ccs-triton7-zk-v3 / JOYZK003, distinct from public execution and public state formats. Its capabilities retain concrete static-shape/work limits, full public state tables, and no live synchronization/deployment claim.
• Neptune transaction and NativeCurrency SDK operations use pinned0.15.1 HardforkGamma policies and require caller-authenticated commitments. The old handwritten os.neptune.proof module remains excluded. Official recursive verification is a replacement API, not rehabilitation of that old module.
• Both warriors advertise deploy=false. Neptune dry-run describes an artifact; it is not live deployment. CPU proving remains the CLI runtime; mining GPU feature selection does not automatically select a GPU prover.
• Complete sibling source archives are the supported packaging route under review. Resource files outside individual crate roots and workspace-root vendor patches remain registry publication blockers. See the exact manifest closure in ../../trident/audit/release-version-closure.md; no registry install or publish success is inferred from workspace tests.

Documentation discrepancies initially reported to the parent (all four subsequently corrected by the parent; no documentation edits made here):

1. docs/explanation/architecture.md says schema/compiler API1 instead of schema1/compiler API2.
2. README's baseline paragraph still says the collection lacks reference coverage; execution now passed133/133 fixtures and43/43 baselines. This does not imply all fixtures have completed proof benchmarks.
3. README's final paragraph must distinguish the excluded old proof module from the now implemented fixed transaction/NativeCurrency recursive SDK.
4. docs/explanation/gpu-backend.md says GPU is on by default; the current CLI uses CPU proving. Feature availability and selected backend must be separate.

This review does not assert universal private execution, private databases, live Neptune state admission, registry publication, or unsupported platform runtime validation.