maestro-guard
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the user via $ARGUMENTS to populate the paths array in .workflow/config.json. This configuration is subsequently used by the workflow-guard hook to enforce write boundaries on the agent's operations.
- Ingestion points: The $ARGUMENTS variable is used to extract paths for the allow and deny subcommands in SKILL.md.
- Boundary markers: The skill does not explicitly use delimiters when interpolating the user-provided paths into the JSON structure, though it performs normalization.
- Capability inventory: The skill uses the Write tool to modify the project's .workflow/config.json and the Read tool to inspect .claude/settings.json.
- Sanitization: Path normalization to forward slashes and deduplication are performed before writing to the configuration file.
Audit Metadata