The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and installs the agent-browser skill from Vercel Labs' official GitHub repository (https://github.com/vercel-labs/agent-browser). This is a trusted dependency used for performing automated web research.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from various external sources including competitor websites, Google Search results, and social forums like Reddit and Quora, creating a surface for indirect prompt injection.\n
Ingestion points: The skill uses agent-browser to extract text from multiple external URLs during the Auto-Discovery phase as defined in references/auto-discovery.md.\n
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to isolate external content from the agent's internal reasoning.\n
Capability inventory: The skill has the ability to execute shell commands (via the browser tool) and write files to the local brands/ directory.\n
Sanitization: There is no evidence of filtering or sanitizing the external content before it is processed by the LLM for synthesis.

marketing-sostac