The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external websites (Google News, LinkedIn, Connectively) using the agent-browser tool without explicit sanitization or boundary markers to isolate potential malicious instructions.
Ingestion points: Scrapes content from various search engines and professional networks via agent-browser as specified in the Research Mode section of SKILL.md.
Boundary markers: No specific delimiters or system instructions are provided to the agent to disregard instructions found within the scraped content.
Capability inventory: The agent has permissions to write files to the local directory structure (./brands/) and navigate the web.
Sanitization: No sanitization or validation of external content is performed before it is added to the agent's context.
[EXTERNAL_DOWNLOADS]: Fetches the agent-browser dependency from Vercel Labs' official GitHub repository during the setup process, which is a trusted source.
[COMMAND_EXECUTION]: Provides instructions to execute agent-browser, npm, and npx commands for research tasks and tool installation (see SKILL.md and references/shared-patterns.md).

marketing-pr