marketing-pipeline-share-content-automation

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It retrieves content from external, third-party sources (such as TechCrunch, X/Twitter, and LinkedIn) through the ResearchService and interpolates this untrusted data directly into the system prompts for Claude and OpenAI models.
  • Ingestion points: Untrusted data enters the context via the ResearchService and is processed by the /api/generate endpoint as shown in the SKILL.md examples.
  • Boundary markers: The prompt templates, such as Create a ${format} post... Research data: ${JSON.stringify(research)}, lack explicit delimiters or instructions for the AI to ignore any malicious commands that might be embedded in the fetched research data.
  • Capability inventory: The AI's output is used to drive the content generation and video rendering pipeline.
  • Sanitization: The provided code snippets do not include logic for sanitizing or filtering the external content before it is submitted to the language models.
  • [EXTERNAL_DOWNLOADS]: The skill instructions direct users to clone a repository from GitHub (https://github.com/pennydinh/marketing-pineline-share.git) and install various dependencies from public registries using standard tools like npm, yarn, or pnpm.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 04:40 PM
Security Audit — agent-trust-hub — marketing-pipeline-share-content-automation