nyc-dob-violations
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from government databases and displays it to the user, which presents a surface for indirect prompt injection if the source data were to contain malicious instructions.
- Ingestion points: JSON responses from multiple datasets on
data.cityofnewyork.us(Datasets3h2n-5cm9,6bgk-3dad, andsjhj-bc8q) as described inSKILL.mdStep 3. - Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore' instructions when handling the external API data.
- Capability inventory: The skill has access to
WebFetchfor network requests andWritefor file system operations, which could be targeted by an injection attack. - Sanitization: The skill does not include specific sanitization or validation routines for text fields retrieved from the API before they are rendered in Markdown.
- [EXTERNAL_DOWNLOADS]: The skill fetches property metadata and violation records from the NYC Open Data (Socrata) platform via official API endpoints at
data.cityofnewyork.us.
Audit Metadata