The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill processes untrusted data scraped from YouTube which could contain malicious instructions designed to influence the agent's analysis or output.
Ingestion points: Video metadata including titles, descriptions, and tags are collected via Apify actors as described in SKILL.md (Step 2) and references/youtube-scraping-guide.md.
Boundary markers: No explicit delimiters (e.g., XML tags or clear separators) or 'ignore embedded instructions' directives are specified for the channel-analyzer sub-agents when they receive scraped data.
Capability inventory: The skill has file-writing capabilities (creating niche-analysis.json and niche-report.md) and access to the WebSearch tool, which could be misused if an injection is successful.
Sanitization: There is no evidence of filtering or sanitizing the external data before it is passed to analysis prompts.

yt-research