server-checkup
Warn
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed for autonomous server maintenance and relies heavily on the use of
sudoto perform system-level changes, such as modifying/etc/ssh/sshd_config, firewall rules, and system service configurations. - [COMMAND_EXECUTION]: Includes specific instructions and a "Quick Reference" for configuring passwordless sudo by writing to
/etc/sudoers.d/, which significantly reduces the authentication hurdles for privileged operations and increases the impact of potential misuse. - [PROMPT_INJECTION]: 'Phase 0' (Context Discovery) instructs the agent to read local markdown files from the server filesystem (e.g.,
server.md,.claude/CLAUDE.md) to determine expected configurations and maintenance history. This creates an indirect prompt injection surface where a malicious file placed on the server could influence the agent's behavior. - Ingestion points: Reads local
.mdfiles and directory listings to extract server-specific context in Phase 0. - Boundary markers: Absent; no instructions are provided to the agent to use delimiters or treat the file content as untrusted data.
- Capability inventory: The skill possesses high-privilege capabilities including software installation/removal and root-level configuration modification.
- Sanitization: Absent; the agent is instructed to directly incorporate findings from these files into its validation and report generation logic.
Audit Metadata