dbs-deconstruct
Pass
Audited by Gen Agent Trust Hub on May 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, external downloads, or privilege escalation attempts were detected. The skill is entirely instructional and does not perform any system actions or network requests.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is designed to ingest and process untrusted user-provided strings for analysis.
- Ingestion points: User-provided concepts are received in Phase 1 of the deconstruction workflow in SKILL.md.
- Boundary markers: Absent; the skill does not use delimiters to isolate user input from its internal logic.
- Capability inventory: None; the skill's functionality is limited to text generation, and it defines no scripts or external tool calls.
- Sanitization: No sanitization or validation of user input is implemented.
Audit Metadata