The Agent Skills Directory

[SAFE]: The skill is a professional developer toolset designed for a specific repository. It includes extensive documentation on handling common pitfalls, licensing, and testing. It incorporates multiple layers of security, including sandboxing and integrity checks.
[COMMAND_EXECUTION]: The instructions involve the use of standard development and automation tools, including git, docker, node, hugo, pytest, and act. These are used appropriately within the scope of managing a local website repository and its associated CI/CD pipelines.
[EXTERNAL_DOWNLOADS]: The skill references downloading official binaries and Alpine Linux packages for building playground engines. It mitigates supply chain risks by explicitly requiring the verification of SHA256 checksums and signatures before inclusion in the build.
[INDIRECT_PROMPT_INJECTION]: The skill processes external tool documentation and metadata during the 'Intake' and 'Feasibility' phases. The potential for indirect injection is addressed by the requirement to use the ai-sanitize tool on generated articles and the inclusion of manual review checkpoints.
[DYNAMIC_EXECUTION]: The core function of the skill is to build and run executable programs. Security is maintained by isolating these programs within restricted WASM module workers or v86 guest environments with disabled networking and restricted kernel configurations.

create-practice-playground