handle-complaint
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to ingest and process text from external, untrusted sources, which introduces a potential consideration for indirect prompt injection if a complaint contains malicious instructions.
- Ingestion Points: The skill retrieves content from external customer communications, including Gmail threads, HubSpot tickets, or text pasted directly by the owner as described in
SKILL.md. - Boundary Markers: There are no explicit delimiters or specific instructions to isolate incoming customer content or warn the agent to ignore any embedded directives.
- Capability Inventory: The execution environment allows access to tools such as
Read,WebFetch, andBashto perform lookups across platforms like HubSpot and PayPal. - Sanitization: The workflow does not explicitly describe any pre-processing, filtering, or validation steps to sanitize the retrieved content before it is evaluated by the agent.
Audit Metadata