abm-strategy

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill mentions the @coldiq/mcp package, which is hosted on the public npm registry.
  • [REMOTE_CODE_EXECUTION]: Includes an instruction to use npx -y @coldiq/mcp@latest to execute the vendor's MCP server directly from the registry. This is a standard deployment method for the vendor's tools.
  • [COMMAND_EXECUTION]: Provides a terminal command for users to manually initialize the Cold-IQ MCP environment.
  • [DATA_EXFILTRATION]: Interfaces with the vendor's API endpoint at api.coldiq.com to perform business-related data enrichment tasks, which is part of the skill's intended functionality.
  • [PROMPT_INJECTION]: The skill processes potentially untrusted user queries and reads from internal documentation to provide its services.
  • Ingestion points: User input and internal files in the references/ directory.
  • Boundary markers: No specific delimiters are employed to isolate external data from instructions.
  • Capability inventory: Access to vendor MCP tools for data retrieval and company enrichment.
  • Sanitization: No sanitization mechanisms are defined for the data processed from reference files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 06:37 PM
Security Audit — agent-trust-hub — abm-strategy