The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a 'Live Website URL Audit' workflow (Context C in shared-patterns.md) which ingests untrusted data from external websites into the agent's context. This creates a surface for indirect prompt injection attacks where malicious instructions hidden on a website could influence the agent's behavior.
Ingestion Points: Live URL content fetched via agent-browser or WebFetch tools as described in shared-patterns.md.
Boundary Markers: No explicit delimiters or warnings to ignore embedded instructions are defined for the website auditing process.
Capability Inventory: The skill has the capability to write files to the project directory (SKILL.md path resolution) and execute browser-based research tools.
Sanitization: No sanitization or filtering logic is specified for the external content before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading and installing external tools and skills required for its research capabilities.
Evidence: shared-patterns.md includes the command npx skills add https://github.com/vercel-labs/agent-browser and npm install -g agent-browser.
Note: The specified source is a well-known technology organization's public repository.

paw-mkt-analytics