ds-paid-audit

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes a restricted Bash tool to execute a local Python script (ds_utils.py) for campaign data processing. This behavior is an intended function of the skill and is constrained by the platform's tool access policies to specific script naming patterns.\n- [PROMPT_INJECTION]: The skill demonstrates an attack surface for indirect prompt injection by processing external campaign data.\n
  • Ingestion points: Marketing metrics and campaign metadata are ingested via MCP tools or manually provided data files in CSV, TSV, or JSON formats.\n
  • Boundary markers: The skill does not implement specific delimiters or explicit safety instructions to isolate ingested campaign data from the agent's operational logic.\n
  • Capability inventory: The agent can execute restricted shell commands and use specialized data retrieval tools to interact with advertising platforms.\n
  • Sanitization: The instructions do not specify any validation, filtering, or sanitization of campaign names or data fields prior to their processing by the utility scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 03:40 AM