The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions require cloning an external GitHub repository (github.com/pennydinh/marketing-pineline-share.git) and installing dependencies from public registries like npm or yarn. This introduces a dependency on external code that should be audited by the user.
[COMMAND_EXECUTION]: The skill utilizes various terminal commands for setup and execution, including npm install, npm run dev, and npm run remotion:render. These commands execute local scripts and binary dependencies within the user's environment.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of crawling external, untrusted web sources (TechCrunch, Twitter/X, LinkedIn) and passing that content directly to LLMs for generation.
Ingestion points: The crawlNewsSource function retrieves data from external URLs which is then used as researchData in generateContent within SKILL.md.
Boundary markers: The provided examples do not include explicit prompt delimiters or instructions to ignore embedded commands within the research data.
Capability inventory: The skill environment contains high-value secrets (OpenAI, Anthropic, and AWS keys) and has the capability to perform network requests and write files to the disk (renderMedia).
Sanitization: There is no evidence of sanitization or filtering of the crawled content before it is interpolated into the AI generation prompts.

ai-content-pipeline-automation