add-report
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill guides the creation of parsers for external data formats (JSONL, CSV, JSON) located in
~/.claude/projects/. This represents a surface for indirect prompt injection where malicious content embedded in these local data files could potentially influence the agent's behavior or cause it to render harmful content when processing or displaying report cards. - Ingestion points: File parsing logic for
~/.claude/projects/*.jsonland other formats in the Infrastructure layer. - Boundary markers: The instructions do not specify the use of delimiters or boundary markers to isolate ingested data from the agent's instructions.
- Capability inventory: The design involves reading local filesystem content and rendering it into SwiftUI components.
- Sanitization: The implementation guidance focuses on formatting metrics (currency, decimals) but lacks instructions for sanitizing or escaping string data extracted from the files for display.
Audit Metadata