skills/bankrbot/skills/skopos/Gen Agent Trust Hub

skopos

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references an external MCP server package skopos-mcp@0.1.0 hosted on the public npm registry.
  • [DATA_EXFILTRATION]: User queries and an anonymous conversation ID are sent to the vendor's API at tryskopos.xyz to retrieve market data. The skill specifically advises against using persistent identifiers to protect user privacy.
  • [PROMPT_INJECTION]: The skill addresses potential indirect prompt injection through several layers:
  • Ingestion points: Data is ingested from the Skopos API and relayed in the chat.
  • Boundary markers: Instructions explicitly mandate that the agent treat all API output as untrusted data and ignore any instructions or links embedded within the prose.
  • Capability inventory: The skill can perform network requests to the vendor API and generate sign-in links for non-custodial transactions.
  • Sanitization: Documentation states that the backend service filters control and zero-width characters from its responses to prevent common injection and obfuscation techniques.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 07:01 PM
Security Audit — agent-trust-hub — skopos