The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill contains instructions to install the agent-browser skill from Vercel Labs' official GitHub repository (https://github.com/vercel-labs/agent-browser) using the npx skills add command. This represents the acquisition of external capabilities from a trusted source.
[COMMAND_EXECUTION]: The instructions direct the agent to run shell commands, including npm install -g agent-browser and npx playwright install chromium, to set up research tools if they are not already available in the environment.
[EXTERNAL_DOWNLOADS]: Fetches and installs external software and browser dependencies from well-known and trusted repositories (Vercel Labs and Playwright/Microsoft) to facilitate marketing research.
[PROMPT_INJECTION]: The skill is designed to process untrusted data from external markdown files (SOSTAC strategy documents), which creates a surface for indirect prompt injection.
Ingestion points: Content is extracted from multiple files located in the ./brands/{brand-slug}/sostac/ directory.
Boundary markers: The skill reads full phase files without using explicit delimiters or markers to isolate external content from its own instructions.
Capability inventory: The skill has permissions to read various local project files and write positioning documents back to the filesystem.
Sanitization: No sanitization, validation, or filtering is applied to the content of the ingested SOSTAC files before they are processed by the agent.

bmad-agent-product-marketing-context