dev-env-setup
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill's audit scripts execute various shell commands within the target project directory, such as 'mise run test', 'mise run serve', and Ruby-based linters (e.g., 'debride', 'flog', 'flay'). These are used to verify the project's runtime compliance and performance characteristics.\n- [DATA_EXFILTRATION]: The compliance checker reads local '.env' files to verify that they are correctly ignored by Git and that their keys match the project's template. It also performs a broad scan of the project directory using 'gitleaks' to identify hardcoded secrets. It includes built-in protection by using the '--redact' flag to prevent secret values from being exposed in logs or terminal output.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it executes arbitrary commands defined in the target project's local configuration files (e.g., 'mise.toml', 'Rakefile'). An attacker could craft a malicious repository that, when audited by this skill, executes arbitrary shell commands through tasks defined in the configuration.\n
- Ingestion points: The target project directory and its local configuration files like 'mise.toml', 'hk.pkl', or 'Rakefile'.\n
- Boundary markers: None identified. The skill directly executes tasks defined in project files as part of its audit.\n
- Capability inventory: Spawning subprocesses via fish and Ruby to run project-defined tasks ('test', 'serve') and dependencies ('bundle exec').\n
- Sanitization: No validation or filtering is applied to the commands defined in the audited project's configuration before they are executed.
Audit Metadata