The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install agent-browser and playwright components. These are well-known technology tools used for the skill's core browser-based research functionality, and shared patterns in the codebase reference these tools from a trusted organization's repository.
[COMMAND_EXECUTION]: The skill makes extensive use of a browser CLI tool to perform web research tasks, such as opening URLs, waiting for network idle, and extracting text or screenshots. These operations are localized to the research session and follow the skill's primary objective of competitive intelligence gathering.
[PROMPT_INJECTION]: The skill contains a potential surface for indirect prompt injection through its data ingestion practices. 1. Ingestion points: The auto-discovery feature scrapes raw text content from external, user-controllable sources such as Reddit, Quora, and G2 reviews. 2. Boundary markers: No specific delimiters or 'ignore embedded instructions' warnings are documented in the processing instructions for this scraped content. 3. Capability inventory: The agent has the ability to write files to the local directory (saving reports) and perform network requests via the browser tool. 4. Sanitization: The skill does not implement specific sanitization or validation of the externally sourced content before it is processed by the AI agent.

bmad-agent-marketing-sostac