The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx @open-slide/cli to download and execute code from the NPM registry. This package is not from a trusted vendor or well-known service, posing a risk of executing unverified code.
[EXTERNAL_DOWNLOADS]: The skill references the open-slide framework at github.com/1weiho/open-slide and fetches dependencies from the NPM registry during initialization.
[COMMAND_EXECUTION]: The skill executes shell commands to scaffold the presentation workspace (npx) and run a local development server (pnpm run dev, npm run dev).
[PROMPT_INJECTION]: The skill directs the agent to load and follow rules from AGENTS.md and skill files in .agents/skills. Since these files are provided by an external third-party tool, this represents an indirect prompt injection surface.
Ingestion points: AGENTS.md and .agents/skills/*.md in the generated workspace.
Boundary markers: None; the agent is instructed to adopt the rules found in these files.
Capability inventory: The skill can read files, execute shell commands, and run network services.
Sanitization: No validation or sanitization is applied to the instructions read from the external workspace.

creating-slides