The Agent Skills Directory

[SAFE]: The skill provides comprehensive and high-quality instructions for auditing Rust codebases, emphasizing read-only operations and the use of repository-defined diagnostic commands. It includes rigorous guidance on verifying unsafe Rust, memory retention, and database interactions without introducing malicious behavior.- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute standard Rust development tools such as cargo check, cargo clippy, and cargo test. These operations are expected within the context of a code audit and are restricted to non-mutating, repository-defined commands.- [INDIRECT_PROMPT_INJECTION]: The skill presents an ingestion surface for untrusted data by processing repository manifests and documentation. 1. Ingestion points: The agent reads Cargo.toml, Rust source files, and guidance documents like AGENTS.md or CLAUDE.md. 2. Boundary markers: The instructions do not define explicit delimiters to separate untrusted repository data from internal instructions. 3. Capability inventory: The skill allows for local code execution through cargo test and cargo bench as part of the audit process. 4. Sanitization: No specific sanitization or filtering logic is prescribed for the files read from the repository.

audit-rust