The Agent Skills Directory

[Command Execution from External Repositories]: The skill instructs the agent to clone GitHub repositories (e.g., github.com/testcontainers/{REPO_NAME}) and execute build and test commands like mvn test, go test, npm test, and pytest on the cloned content. While targeting a well-known organization, executing code from external sources is a security consideration as it runs scripts and installs dependencies defined in those repositories.
[Privilege Escalation Surface via Docker Socket Mount]: To verify that Testcontainers guides work, the skill mounts the host's Docker socket (/var/run/docker.sock) into containers that execute the cloned test code. This configuration provides the container with significant control over the host's Docker daemon. This is a standard requirement for Testcontainers functionality but warrants review when used to test code from external repositories.
[Potential for Indirect Prompt Injection]: The skill processes AsciiDoc files and source code from external repositories to generate documentation. If an external guide repo were to contain malicious instructions embedded in its text, the agent might interpret them as tasks. The skill lacks explicit boundary markers or instructions to ignore commands found within the guide content.
[Execution of External Build Tools]: The skill uses docker run to execute various language toolchains (Maven, Go, Node.js, Python, .NET) and documentation tools like rumdl. This involves downloading and running external packages and dependencies at runtime during the verification and linting phases.

testcontainers-guide-migrator