The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation suggests executing commands such as cargo build, cargo run, and cargo test. These operations involve an inherent risk of executing arbitrary code if the target Rust project is untrusted, specifically through malicious build.rs scripts or compromised test cases.
[COMMAND_EXECUTION]: The skill contains installation instructions for Linux package managers that utilize sudo (e.g., sudo apt install rust-analyzer). An automated agent following these instructions would be performing privilege escalation.
[NO_CODE]: The skill does not contain any executable scripts (e.g., Python or Bash files). It is comprised entirely of Markdown instructions and metadata, reducing the risk of direct malicious code execution from the skill itself.
[COMMAND_EXECUTION]: Indirect Prompt Injection Surface Analysis:
Ingestion points: The agent reads Rust source code (.rs files) and project configuration (Cargo.toml) from the local environment.
Boundary markers: None are defined; instructions do not advise the agent to ignore embedded commands in project files.
Capability inventory: The skill encourages the use of cargo commands which spawn subprocesses capable of running arbitrary code defined in the project structure.
Sanitization: There is no mention of sanitizing or validating project files before execution.

rust-analyzer-lsp