pricing-strategy
Warn
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple local shell commands and binaries to manage state and configurations. Evidence: The preamble and post-output sections of SKILL.md invoke local utilities such as rmbc-update-check, rmbc-analytics, and rmbc-workspace.
- [PROMPT_INJECTION]: The skill is vulnerable to command injection by interpolating untrusted user input into a shell execution. 1. Ingestion points: User responses captured via the AskUserQuestion tool. 2. Boundary markers: Absent. 3. Capability inventory: Execution of local binaries through the shell. 4. Sanitization: Absent. Evidence: The instruction /bin/rmbc-workspace active "<user's answer>" in SKILL.md.
- [EXTERNAL_DOWNLOADS]: The skill references and opens an external URL. Evidence: The welcome sequence in SKILL.md uses the open command for a YouTube video link, which is a well-known service.
- [DATA_EXFILTRATION]: The skill includes a telemetry logging feature that executes a local script likely performing network operations. Evidence: The preamble in SKILL.md executes rmbc-analytics with specific usage metadata. The skill provides a clear opt-out path for the user.
Audit Metadata