Skills
Navigate
skills/finpeakinc/frevana-skills/claude-sonnet-5/Gen Agent Trust Hub

claude-sonnet-5

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes local scripts and Python code to manage API interactions.\n
  • Evidence: scripts/create_response.sh uses bash and inline python3 scripts for JSON generation and response parsing.\n- [EXTERNAL_DOWNLOADS]: Connects to an external service to fetch AI-generated responses.\n
  • Evidence: Sends HTTPS POST requests to https://ai-factory.frevana.com/openrouter/v1/responses via curl.\n- [INDIRECT_PROMPT_INJECTION]: Processes untrusted data from user inputs that are forwarded to an external LLM.\n
  • Ingestion points: Arguments including --input, --input-file, --instructions, and --instructions-file in scripts/create_response.sh.\n
  • Boundary markers: No explicit delimiters are added to separate user input from system instructions in the prompt construction.\n
  • Capability inventory: The skill has network access via curl and file-writing capabilities for session files and output capture.\n
  • Sanitization: Employs Python's json.dump to ensure all user-provided strings are correctly escaped and formatted as valid JSON before transmission.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 07:30 PM