The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an inherent attack surface for indirect prompt injection because its core functionality involves automated browsing and text extraction from third-party websites (such as Reddit, Quora, and G2) to generate strategic insights.
Ingestion points: Data enters the context from external URLs via browser automation commands in references/auto-discovery.md.
Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions for content retrieved from the web.
Capability inventory: The skill uses agent-browser for networking, npm for tool installation, and standard file operations (ls, cat) for workspace management.
Sanitization: Web content is processed for analysis without explicit sanitization or filtering steps.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install agent-browser and Playwright Chromium via npm. These downloads are performed from the official npm registry and originate from a trusted organization (Vercel Labs).
[COMMAND_EXECUTION]: The skill uses shell commands for environment setup and browser automation. These operations are well-documented and essential for the automated research-first methodology the skill provides.
[SAFE]: The skill includes proactive security guidance in references/shared-patterns.md, warning users to use dedicated browser profiles for automation and to protect sensitive session tokens saved in authentication state files.

paw-mkt-sostac