job-application-operator
Pass
Audited by Gen Agent Trust Hub on May 22, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses sensitive personal information stored in local configuration files (e.g.,
~/.config/AgentDesk/personal-profile.yaml). This access is necessary for the skill's core purpose of filling job application forms and is mitigated by instructions to never commit this data to version control. - [PROMPT_INJECTION]: The skill processes untrusted content from external job application websites, creating a surface for indirect prompt injection.
- Ingestion points: Web content from job application URLs processed during the application loop.
- Boundary markers: No explicit delimiters are defined for separating external web content from instructions.
- Capability inventory: File system read/write for local configuration and logs, and browser automation using tools like Playwright or
agent-browser. - Sanitization: Mitigation is provided through hard rules requiring human confirmation before final submission and instructions to stop and ask for review when encountering sensitive or unknown fields.
Audit Metadata