The Agent Skills Directory

[SAFE]: The skill provides instructions for utilizing the mcporter utility to manage and call MCP servers on demand. This approach is described as a strategy to optimize resources in small containers by avoiding persistent background processes.
[COMMAND_EXECUTION]: The skill documents the use of shell commands to list servers and call tools via the mcporter CLI. These operations are essential to the skill's primary function and include explicit instructions to avoid inspecting credential files.
[INDIRECT_PROMPT_INJECTION]: The skill incorporates tools like web-search-prime and web-reader which ingest content from external websites and search results. While this creates a surface for potential indirect prompt injection, it is the standard and intended behavior for web-enabled agent capabilities.
Ingestion points: External data enters the agent context through web-reader.webReader and web-search-prime.web_search_prime calls in SKILL.md.
Boundary markers: None explicitly defined in the provided instruction set.
Capability inventory: The agent has access to the shell to run mcporter commands.
Sanitization: No specific sanitization or filtering logic is described for the content retrieved from remote URLs or search queries.

mcp-skill