The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to clone a codebase from an unverified third-party GitHub repository (https://github.com/pennydinh/marketing-pineline-share.git). This source does not belong to a known trusted organization and is not clearly linked to the skill author's infrastructure, presenting a potential supply chain risk.
[PROMPT_INJECTION]: The content generation module in src/modules/content/generator.ts is vulnerable to indirect prompt injection. It fetches content from external sources (Twitter/X, LinkedIn, TechCrunch) and interpolates this untrusted data directly into the system prompt for the AI model without sanitization or boundary markers.
Ingestion points: The researchData object is stringified and injected into the prompt within buildPrompt().
Boundary markers: None. Data is placed directly into the instruction block.
Capability inventory: The agent has the ability to publish to social media via the Facebook Graph API and interact with a local database via Prisma.
Sanitization: No filtering or escaping is applied to the external content before it is processed by the LLM.
[COMMAND_EXECUTION]: The skill documentation provides various CLI commands for setup and execution, including installation of system dependencies via sudo apt-get. While these are part of the stated functionality, they involve high-privilege operations that would execute code from the unverified repository.

marketing-pipeline-automation