cold-outreach
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No instructions designed to bypass safety filters or override agent behavior were detected. The skill uses standard instructional language.
- Data Exposure & Exfiltration (SAFE): No hardcoded credentials, sensitive file paths, or network operations are present. The skill only handles text provided by the user within the chat context.
- Obfuscation (SAFE): The content is clear and uses standard Markdown. No Base64, zero-width characters, or hidden encodings were found.
- Unverifiable Dependencies & Remote Code Execution (SAFE): No external packages or remote scripts are referenced or executed. The skill is entirely self-contained within the Markdown files.
- Privilege Escalation (SAFE): No commands related to administrative privileges or system-level modifications (e.g., sudo, chmod) are included.
- Persistence Mechanisms (SAFE): No attempts to modify startup scripts, crontabs, or registry keys were found.
- Indirect Prompt Injection (LOW): The skill processes user-provided context such as job descriptions and company websites. While this is an ingestion point for untrusted data, the skill lacks the capabilities (network access, file writing, or command execution) to be exploited via this vector. The risk is limited to the generation of unintended text in the email draft.
Audit Metadata