file-management

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides methods to read file content and search snippets, which could allow malicious instructions from untrusted files to enter the agent's context. Ingestion points: SKILL.md (endpoints /v1/files/{fileId}/content and /v1/search). Boundary markers: Absent. Capability inventory: File CRUD and metadata updates via data.spuree.com. Sanitization: Raw file content is ingested without validation; search results are HTML-escaped except for mark tags.
  • [DYNAMIC_EXECUTION]: The skill includes a shell command template that uses a Python one-liner for local CRC32 calculation. Evidence: python3 -c script for checksum computation in SKILL.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 04:01 PM
Security Audit — agent-trust-hub — file-management