The Agent Skills Directory

[COMMAND_EXECUTION]: The mcporter CLI includes a --stdio flag that enables the execution of arbitrary shell commands (e.g., bun run ./server.ts). This allows the agent to spawn local processes based on user or server-provided strings.
[EXTERNAL_DOWNLOADS]: The skill's metadata specifies the installation of the mcporter package from the Node.js registry. Additionally, the call command supports interacting with remote HTTP MCP servers, leading to external network requests.
[REMOTE_CODE_EXECUTION]: The tool includes features for code generation (generate-cli and emit-ts), which create executable CLI wrappers or TypeScript types based on remote server schemas. This could lead to the creation and execution of malicious code if a remote server provides a compromised schema.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from remote MCP servers that are then processed by the agent.
Ingestion points: Data enters the agent's context through mcporter list <server> --schema and the results of mcporter call <server.tool>.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are documented for the output of these tools.
Capability inventory: The tool can execute shell commands via --stdio, perform network operations via HTTP tools, and write files via the generate-cli command.
Sanitization: There is no evidence of validation or sanitization of the schemas or data returned by external MCP servers before they are used to influence agent behavior or generate code.

mcporter