The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation URL https://jeffallan.github.io/claude-skills/skills/language/rust-engineer/ included in the skill body and metadata is flagged as malicious and blacklisted by automated reputation scanners.
[METADATA_POISONING]: The skill metadata contains references to the blacklisted domain jeffallan.github.io, which may be used to direct users or agents to untrusted external resources.
[INDIRECT_PROMPT_INJECTION]: The skill provides a workflow for processing, linting, and testing user-provided Rust code, creating an attack surface where malicious data could influence agent behavior.
Ingestion points: The skill processes user code blocks, project files, and Cargo manifests (SKILL.md, references/testing.md).
Boundary markers: There are no explicit instructions to treat user-provided code as untrusted or to use isolation when running tests.
Capability inventory: The skill encourages the use of cargo test, cargo clippy, and cargo bench. These commands can execute arbitrary code during the build or test phase via build.rs scripts or procedural macros.
Sanitization: No sanitization or sandboxing procedures are defined for the execution of developer tools on external code.
[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands (cargo test, cargo clippy, etc.) which, when combined with untrusted user code, can lead to local code execution on the user's machine.

rust-engineer