The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the standard use of the sv CLI for project scaffolding and maintenance. It also describes a programmatic API (sv.execute) used by add-on developers to run shell commands within the project environment.
[EXTERNAL_DOWNLOADS]: The instructions describe fetching the sv tool and various add-ons from the npm registry using standard package managers like npm, pnpm, bun, and deno. It includes a specific reference to community add-ons with a clear security disclaimer that these are not audited by the Svelte maintainers.
[PROMPT_INJECTION]: The skill identifies surfaces for indirect prompt injection where user-supplied inputs, such as add-on names or configuration flags, are interpolated into command executions or file modifications.
Ingestion points: CLI arguments and interactive prompt responses during project creation or add-on installation (documented in SKILL.md).
Boundary markers: No specific boundary markers or instruction-following delimiters are documented for these inputs.
Capability inventory: The documented add-on system provides high-tier capabilities including shell command execution (sv.execute), file system modifications (sv.file), and dependency management (sv.dependency) as detailed in references/custom-addon-reference.md.
Sanitization: The documentation does not explicitly detail input sanitization or validation for these arguments.

svelte-cli