The Agent Skills Directory

REMOTE_CODE_EXECUTION (CRITICAL): Remote script from an untrusted source piped directly to shell.\n
Evidence: Automated scanning detected the command curl -sS https://starship.rs/install.sh | sh. This is an extremely dangerous pattern that executes unverified code from the internet, bypassing security reviews and integrity checks.\n- COMMAND_EXECUTION (HIGH): Excessive privileges granted through 'Docker-in-Docker' (dind).\n
Evidence: In templates/devcontainer.json, the feature ghcr.io/devcontainers/features/docker-in-docker:2 is enabled. Providing the container with access to the host's Docker daemon is a major security risk that can be exploited for container escape and host-level compromise.\n- DATA_EXFILTRATION (MEDIUM): Automatic exposure of sensitive environment credential files.\n
Evidence: The configuration in templates/devcontainer.json includes a bind mount for the host's .env file (source=${localWorkspaceFolder}/.env,target=${containerWorkspaceFolder}/.env). This automatically transfers host-side secrets, API keys, and environment variables into the container, increasing the risk of credential theft.\n- REMOTE_CODE_EXECUTION (LOW): Automatic execution of untrusted package installation during startup.\n
Evidence: The postCreateCommand in templates/devcontainer.json executes npm ci upon environment creation. This automatically installs and runs code from third-party registries based on local configuration, posing a supply-chain risk if dependencies are compromised.

development-environment