The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from the external website raspberrypi.org, which could potentially contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Tools such as raspberrypi_org__read_page and unbrowse.scrape (documented in SKILL.md) retrieve live web content.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to ignore embedded commands within the fetched data.
Capability inventory: The skill includes powerful browser automation capabilities via unbrowse.browse.act and unbrowse.run (documented in SKILL.md).
Sanitization: There is no mention of sanitizing or filtering retrieved web content before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill instructions facilitate the download of MCP configuration and metadata from the vendor's official domain.
Evidence: The README.md and SKILL.md provide commands to add MCP servers and skills from https://unbrowse.ai.
[COMMAND_EXECUTION]: The documentation includes shell commands for users to set up the MCP environment.
Evidence: README.md includes claude mcp add and npx skills add commands.

raspberrypi