iterative-refinement
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (MEDIUM): The file
tools/rust.mdinstructs the agent to install several unverified third-party crates globally usingcargo install, includingcargo-tarpaulin,cargo-llvm-cov,cargo-audit,cargo-deny,cargo-criterion,cargo-nextest, andsccache. - [COMMAND_EXECUTION] (LOW): The skill's primary functionality relies on executing a variety of shell-based validation tools and a provided bash script template. While these are standard developer tools, the automated execution environment poses a risk if processing compromised codebases.
- [INDIRECT_PROMPT_INJECTION] (LOW): The refinement workflow ingests data from external tool logs and suggested web search results without sanitization. Evidence Chain: 1. Ingestion points:
SKILL.md(search results) andtools/rust.md(tool logs). 2. Boundary markers: Absent. 3. Capability inventory: Execution ofcargosuite andbashscripts intools/rust.md. 4. Sanitization: Absent. This enables untrusted content to influence the agent's logic during iterative cycles.
Audit Metadata