The Agent Skills Directory

Dynamic Module Loading and Transformation: The logic in src/executor-init.ts performs runtime transformation of JavaScript module code. It rewrites import statements and utilizes dynamic import() with base64-encoded data URLs to load internal shims. This approach, while facilitating compatibility with upstream dependencies, involves the dynamic generation and execution of code.
Remote Resource Fetching: The skill is designed to fetch and process API specifications (OpenAPI, GraphQL) and connect to remote MCP servers. This involves outbound network connections to URLs defined by the user in the skill's configuration.
Host-Side Command Execution: Through its MCP integration, the skill can execute local commands via the stdio transport. This allows the agent to interact with host-side CLI tools, which requires careful management of the commands and arguments passed to the system.
Indirect Input Handling: As a tool-invocation companion, the skill processes definitions and responses from external services. There is a potential consideration for indirect prompt injection if the external data source (like a compromised API spec) contains instructions intended to influence the agent's behavior.

just-bash-executor