Rust adoption-to-diagnosis proof

This page documents the first complete review-first Rust product vertical in SDETKit. The proof composes the existing adoption-surface, FailureVector, SafetyGate, and Doctor report contracts. It does not introduce a Rust-specific control plane.

Operator journey

Rust repository detected
→ Cargo proof surfaces identified from repository evidence
→ explicit cargo-audit evidence reported
→ saved cargo test failure normalized into FailureVector
→ SafetyGate keeps the result review-first
→ Doctor report carries the failure evidence and next human action

The fixture used for this proof is tests/fixtures/adoption_repos/rust_cargo. The saved CI evidence is tests/fixtures/ci_failures/cargo_test/ci_log.txt.

What is detected

SDETKit reads repository files and reports:

• language: rust, from Cargo.toml
• package manager: cargo, from Cargo.toml
• proof command: cargo test
• security surface: cargo_audit, only because the fixture contains an explicit workflow command
• security proof command: cargo audit

Cargo.toml alone does not imply cargo-audit adoption. The security surface is emitted only when explicit workflow, script, or configuration evidence exists.

What is inferred

The adoption surface recommends commands from repository evidence, but does not execute them. Every recommended command keeps auto_run_allowed=false because Cargo commands may build or execute untrusted target code.

The saved Cargo log is classified as:

• ecosystem: rust
• tool: cargo_test
• failure class: test
• affected file: src/lib.rs
• exit code: 101
• local proof command: cargo test
• confidence: high

The log is read as evidence. SDETKit does not install Rust, run Cargo, modify Cargo.lock, or touch the target repository.

What is proven

The integration proof verifies that the normalized Rust FailureVector can be passed into the existing Doctor report contract. Doctor then reports:

• status: review_required
• confidence: high
• roadmap lane: failure_diagnosis
• top failure check: cargo-test
• top failure type: test
• next local proof: cargo test

The same evidence is rendered as deterministic JSON and Markdown artifacts in the test proof.

Authority boundary

The complete vertical explicitly preserves these values:

target_code_execution=false
auto_run_allowed=false
automation_allowed=false
patch_application_allowed=false
security_dismissal_allowed=false
merge_authorized=false
semantic_equivalence_claim=false

This proof demonstrates diagnosis and reporting capability. It does not authorize remediation.

What remains manual

A human operator must still:

1. review the repository and saved failure evidence;
2. decide whether running Cargo commands is appropriate in the target environment;
3. inspect the failing Rust code and test;
4. review cargo-audit findings without automatic dismissal;
5. approve any patch, dependency change, lockfile change, or merge separately.

Reproduce the repository proof

python -m pytest -q \
  tests/test_rust_adoption_vertical.py \
  tests/test_adoption_surface_cargo_audit.py \
  tests/test_cross_ecosystem_failure_vector_adapters.py \
  tests/test_doctor_report_contract.py \
  -o addopts=
python -m mypy src
python -m pre_commit run -a
NO_MKDOCS_2_WARNING=1 python -m mkdocs build --strict

No command above executes the Rust fixture. The fixture and saved log remain read-only product evidence.