gstack-openclaw-ceo-review
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is purely instructional and establishes a reasoning framework for plan evaluation. It does not contain executable code, remote downloads, or hidden payloads.
- [PROMPT_INJECTION]: While the skill processes user-supplied plans (a surface for indirect prompt injection), it contains strong defensive instructions: it explicitly forbids the agent from making code changes or beginning implementation, and mandates that the user remains 100% in control of all decisions. The skill also requires the agent to perform its own threat modeling on the plans it reviews.
- [COMMAND_EXECUTION]: Although the skill has access to the 'Bash' tool, the instructions restrict its use to analysis rather than implementation. No suspicious command sequences or privilege escalation attempts were found.
- [DATA_EXFILTRATION]: There are no network requests, credential access attempts, or exfiltration patterns detected in the instructions.
Audit Metadata