The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use repository-local CLI tools for development tasks, including vp create, vp migrate, vp check, vp test, and vp pack. These operations are performed through the project's package manager (e.g., pnpm exec), ensuring execution is tied to the project's own version-controlled environment.
[EXTERNAL_DOWNLOADS]: The skill references established GitHub Actions (voidzero-dev/setup-vp) and GitLab CI templates for environment setup. These are legitimate resources provided by the toolchain authors for automated pipelines.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing existing repository files (manifests, configs, and source code) to perform migrations. While this is a common attack surface for indirect prompt injection, the skill mitigates this by instructing the agent to perform verification steps, such as vp toolchain --json and manual manifest inspections, rather than relying solely on generated output.
[CREDENTIALS_UNSAFE]: Security best practices are encouraged in the CI/CD documentation, which explicitly recommends isolating dependency caches for secret-bearing jobs to prevent cross-environment data leakage during automated releases.

vite-plus