filesystem-context

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill describes methods for ingesting potentially untrusted data from external tools, terminal sessions, and user preferences into the filesystem, which could be leveraged for indirect prompt injection attacks.
  • Ingestion points: Web search results (Pattern 1), terminal logs (Pattern 5), and user preferences (Pattern 6) in SKILL.md.
  • Boundary markers: The instructions lack specific guidance on using delimiters or safety markers to isolate external data.
  • Capability inventory: The pattern assumes access to file system tools (write_file, read_file, write_yaml, load_yaml) and search utilities (grep, ls).
  • Sanitization: No explicit content validation or sanitization steps are defined for the ingested data.
  • [SAFE]: The external references provided (e.g., LangChain, Anthropic, Cursor) are to well-known and reputable industry sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 09:43 AM
Security Audit — agent-trust-hub — filesystem-context