debugging
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (HIGH): The skill directs the agent to execute commands like 'cargo build' and 'cargo run'. These are high-risk operations because Rust's build system allows 'build.rs' scripts to execute arbitrary code on the host machine during compilation.
- [EXTERNAL_DOWNLOADS] (LOW): The skill suggests downloading and installing tools such as Miri and cargo-flamegraph. While these come from trusted sources within the Rust ecosystem, they represent external dependencies.
- [PROMPT_INJECTION] (HIGH): The skill is susceptible to Indirect Prompt Injection (Category 8) because it ingests untrusted external data (application code, logs, and error messages) and possesses powerful write/execute capabilities without implementing sanitization or boundary markers.
Recommendations
- AI detected serious security threats
Audit Metadata