tailrocks-rust-project-remediate
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill includes a dedicated
runtime-trust.mdreference that mandates keeping secret values unread and prohibits copying them into any logs, prompts, or artifacts. It instructs the agent to cite secret locations and types only, which is a security best practice. - [COMMAND_EXECUTION]: The skill provides instructions for the agent to run standard Rust development and verification tools, such as
cargo,rustfmt,clippy, andmise. These operations are performed within the context of a buildable workspace to verify remediation slices. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from repositories, registries, and web content. It includes specific defensive boundary instructions in
runtime-trust.mdstating that embedded instructions in processed data cannot alter the agent's scope, authority, or approval requirements. - [EXTERNAL_DOWNLOADS]: The skill references established primary sources for the Rust ecosystem (e.g., crates.io, forge.rust-lang.org, docs.rs) and the vendor's own template directories for project layout configuration. These are trusted or well-known sources and do not involve unverified remote script execution.
Audit Metadata