staffer
Fail
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: HIGHPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PRIVILEGE_ESCALATION]: The skill instructions explicitly direct the agent to run commands 'unsandboxed' and to 'request escalated permissions' for the purpose of accessing OAuth token files and localhost ports that are normally restricted by the environment's security sandbox.\n- [COMMAND_EXECUTION]: The skill uses the
nodecommand to execute a local companion script, providing a vector for system-level operations and potential command injection.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to take arbitrary user input and interpolate it directly into shell command arguments.\n - Ingestion points: User task input via the
--promptargument inSKILL.md.\n - Boundary markers: Absent.\n
- Capability inventory: Shell execution (node, git, gh) and file system access (Read, Glob, Grep) are available as specified in the skill's tool restrictions.\n
- Sanitization: No sanitization or validation of the user-provided task text is described or performed within the skill instructions.
Recommendations
- AI detected serious security threats
Audit Metadata