The Agent Skills Directory

Library Security Hardening: During the bundling process, the bundle.mjs script applies specific security patches to the reveal.js library. It "neutralizes" certain sinks, such as window message listeners and certain attribute assignments, which reduces the potential attack surface of the generated standalone HTML.
Filesystem Integrity Protections: The underlying build scripts (build.mjs, create-deck.mjs) include advanced filesystem validation. By using O_NOFOLLOW flags and verifying file identity via device and inode numbers, the skill effectively mitigates risks associated with symbolic link traversal and potential race condition attacks on the local filesystem.
Controlled Dependency Restoration: The skill's documentation and SKILL.md instructions emphasize using npm ci --ignore-scripts for dependency management. This approach is a positive security practice that prevents the automatic execution of potentially untrusted lifecycle scripts (e.g., preinstall) defined in a package's configuration.
Workspace Build Architecture: The build-decks.mjs script utilizes dynamic import() to load and execute bundle.mjs files found within the user's workspace. This is a core component of the "build-all" feature, allowing the skill to run localized bundling logic for different decks. This architecture is documented within the skill's comprehensive security model, which emphasizes its use within trusted workspaces.

hve-slides