The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation instructs the agent to install the @zeeg/dex package from a public npm registry. This is a standard dependency for the tool's core functionality.
[COMMAND_EXECUTION]: The skill relies on executing various dex commands to manage task states, and utilizes standard development tools like npm and git for environment setup and verification. It also suggests running local repository scripts such as node scripts/audit-content.mjs.
[INDIRECT_PROMPT_INJECTION]: The skill features commands like dex plan and dex import that ingest data from external files and GitHub issues, creating a surface for indirect prompt injection. 1. Ingestion points: dex plan (reads local markdown files) and dex import (fetches GitHub issue data). 2. Boundary markers: None explicitly defined to delimit external content from agent instructions during ingestion. 3. Capability inventory: The agent can perform file system writes (.dex/ directory), execute shell commands (dex, npm, git), and make network requests (GitHub/Shortcut API). 4. Sanitization: There is no mention of content sanitization or validation during the import process.

dex