migrate-to-rust
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The skill involves executing several shell commands, including
git clone,cargo,rustc, andllvm-profdata. It also executes a user-defined{baseline_run_command}to benchmark the original legacy code during the discovery phase. - External Code Interaction: The skill clones code from external GitHub repositories. To mitigate risks associated with executing unknown code, the instructions explicitly mandate a 'Stage 0' confirmation gate, where the agent must present the resolved build and run commands to the user for approval before proceeding with compilation or execution.
- Indirect Prompt Injection Surface: Since the skill ingests and analyzes external codebases (including files like READMEs or build scripts), it has an inherent surface for indirect prompt injection. However, the structured workflow and mandatory human-in-the-loop checkpoint for code execution provide significant mitigation for this potential vulnerability.
- Data Handling: The skill records performance metrics and codebase census data in a local JSON file (
migration_telemetry.json) and generates an HTML report. All data processing is performed locally within the workspace environment.
Audit Metadata