asc-apple-ads
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to guide an agent in executing the
ascCLI tool for administrative tasks, including fetching reports, discovering account resources, and performing mutations.\n- [CREDENTIALS_UNSAFE]: The documentation mentions sensitive file paths such as~/.asc/apple-ads-private-key.pemand environment variables for storing OAuth credentials. While these are standard for the operation of theasctool, they represent sensitive data access points within the agent's operating environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external JSON files to build API payloads and search queries, which constitutes a potential attack surface.\n - Ingestion points: The agent reads data from local files provided via the
--fileflag (e.g.,query.json,report.json,paused-campaign.json).\n - Boundary markers: The instructions include safety markers such as requiring the
--confirmflag for mutations and recommending user review of all payloads.\n - Capability inventory: The skill enables the agent to perform write operations (create, update, delete) and make raw API requests to the Apple Ads Platform API.\n
- Sanitization: Mitigation relies on human oversight and explicit confirmation prompts rather than automated input filtering.
Audit Metadata