ucoz-ad-campaign-landing-skill

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands via the agent to perform various tasks. This includes using curl to interact with Yandex Direct, Google Ads, and Yandex Search APIs, as well as executing PowerShell scripts for binary image generation (favicon.ico) and Python one-liners for URL encoding and JSON parsing of API responses.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by processing untrusted data from external sources.
  • Ingestion points: Data enters the agent's context through user-uploaded CSV files (ad content) as described in ADS.md, and through external API results (Yandex Wordstat and SERP data) as described in RESEARCH.md.
  • Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore potentially malicious content embedded in the ingested CSV data or API responses.
  • Capability inventory: The skill possesses significant capabilities, including network access via curl, file system manipulation via ftp_tool (writing CSS/JS/images), and site modification via templates_tool (publishing HTML pages).
  • Sanitization: The instructions do not specify any validation, filtering, or escaping of the ingested data before it is interpolated into HTML templates or API payloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 10:58 AM
Security Audit — agent-trust-hub — ucoz-ad-campaign-landing-skill