astrofy-plan-about-page
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns detected. The skill instructions focus on gathering institutional information for marketing copy and emphasize verifying facts and identifying non-public statements.
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from an interviewee. However, this is evaluated as safe because the skill includes specific instructions to verify all sources, mark content as draft until manual approval, and explicitly forbids inventing data.
- Ingestion points: User-provided interview responses defined in
SKILL.md. - Boundary markers: Instructions to mark copy as 'rascunho' (draft) and distinguish between confirmed facts and opinions.
- Capability inventory: No tool executions or external network operations are triggered by the ingested data; it only produces a text fragment (
type: about). - Sanitization: Mandatory source checks for dates, numbers, and credentials as specified in
references/technical.md.
Audit Metadata