The Agent Skills Directory

[Privilege Escalation]: The instructions include the use of sudo for administrative operations, specifically for installing packages via apt-get. Executing commands with elevated permissions is a security consideration during environment setup.
[Command Execution]: The skill directs the agent to execute a broad range of system commands and development tools (e.g., cmake, dotnet, go, npm, cargo, swift, pip) to build, test, and lint code, which is essential for its primary function.
[External Downloads]: The skill references external source code from GitHub repositories (e.g., for googletest and Catch2) and utilizes various package registries to manage testing dependencies. These sources are established and well-known in the developer community.
[Dependency Management]: The guidance includes instructions for installing PowerShell modules with flags that bypass publisher verification (e.g., SkipPublisherCheck), which may affect the verification of package integrity.
[Indirect Prompt Injection]: The skill's workflow involves reading source code and project configuration files to generate unit tests, which introduces an attack surface for indirect prompt injection.
Ingestion points: Project source files (src/, lib/), build manifests (e.g., package.json, Cargo.toml, pom.xml), and existing test suites.
Boundary markers: The instructions do not define specific delimiters or headers to separate untrusted code content from the agent's internal logic.
Capability inventory: The agent is instructed to perform file system operations, execute shell commands, and interact with network-based package managers.
Sanitization: No explicit sanitization or filtering of the analyzed source code is described in the provided instructions.

code-testing-extensions