atmee-product

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to retrieve live data (pricing, account status, avatar lists) from the atmee MCP server to answer user queries, which creates a surface for instructions embedded in that data to influence agent behavior.
  • Ingestion points: Untrusted data enters the agent context through MCP tool calls such as get_account, get_balance, and list_avatars in SKILL.md.
  • Boundary markers: The instructions lack explicit delimiters or "ignore embedded instructions" warnings when handling the data returned from the API.
  • Capability inventory: The skill uses the atmee MCP tool (https://api.atmanity.us/mcp) as defined in openai.yaml. No shell execution or local file-write capabilities were found in the analyzed files.
  • Sanitization: There is no documentation of sanitization, escaping, or schema validation for the data ingested from the remote server.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 06:16 PM