tailrocks-rust-project-audit
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill operates in a read-only mode, explicitly forbidding file modifications or tool installations during the audit process. It utilizes well-known ecosystem tools (such as
cargo-denyandcargo-audit) from established registries likecrates.ioto perform policy and vulnerability checks. - [INDIRECT_PROMPT_INJECTION]: The skill analyzes external codebases, which constitutes an ingestion point for untrusted instructions, but it implements specific defenses against indirect injection.
- Ingestion points: Target Rust repository files, module layouts, and dependency manifests as described in
SKILL.md. - Boundary markers:
references/runtime-trust.mdprovides explicit instructions to treat repository content as untrusted data that cannot override agent instructions or authority. - Capability inventory: Utilizes
cargoaudit and linting tools for structural and policy analysis without mutation. - Sanitization: Mandates pre- and post-audit repository byte hashing and utilizes frozen/locked dependency resolution to ensure data integrity.
Audit Metadata