The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run standard Rust developer tools including cargo fmt, cargo clippy, cargo test, cargo miri, and cargo semver-checks. While these can execute code, the skill explicitly warns the user to only run them in sandboxed environments after manual inspection and approval.
[INDIRECT_PROMPT_INJECTION]: This skill has an inherent attack surface for indirect prompt injection because its core function is to process and analyze untrusted third-party Rust source code and pull requests.
Ingestion points: The skill reads Rust source files (.rs), dependency manifests (Cargo.toml), and pull request diffs provided in the agent's context.
Boundary markers: The skill does not define specific text delimiters for the processed code, but it provides explicit instructional boundaries, advising the agent to prioritize manual reading over execution when code trust is low.
Capability inventory: The agent has the capability to execute shell commands via the cargo tool suite, which can be exploited by malicious build scripts (build.rs) or procedural macros in the reviewed code.
Sanitization: The skill implements a robust safety protocol, requiring the agent to perform manual analysis of all executable build components, obtain user consent before tool execution, and utilize sandboxing with no network access or secrets.

rust-review