The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches documentation from docs.uniac.ai and installs the uniac CLI tool from the public NPM registry.
[COMMAND_EXECUTION]: The skill uses the uniac CLI to perform core operations, including uniac plan for validation, uniac project create for resource provisioning, and uniac deploy for application lifecycle management.
[REMOTE_CODE_EXECUTION]: The instructions suggest using npx to execute the vendor-provided @uniac/cli tool directly from the network, which is a standard execution pattern for this platform.
[INDIRECT_PROMPT_INJECTION]: The skill ingests application metadata and instructions from uniac.yaml files provided by the user.
Ingestion points: uniac.yaml configuration files and project-specific manifests referenced in SKILL.md and references/composition.md.
Boundary markers: Not explicitly defined in the instructional text.
Capability inventory: File system access (reading configuration, writing deployment metadata to .uniac/deploy.json), network access (pushing Docker images to the Uniac registry), and container orchestration via the local Docker daemon.
Sanitization: The skill relies on the uniac plan command to perform schema validation and reference checks on manifests before any deployment actions are taken.

uniac