The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill directs the agent to execute shell commands that download and run external Node.js packages from a package registry without version pinning.
Evidence in SKILL.md: The command vp dlx --package comark-linter --package comark-ui lint-comark lesson.md fetches and executes packages at runtime.
[COMMAND_EXECUTION]: The agent is explicitly instructed to run shell commands in the local environment to lint generated files.
Evidence in SKILL.md: Section "5. Check the render" contains a command to lint the output using a CLI tool.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from conversation history and external technical sources to generate content that is then handled by execution tools.
Ingestion points: Conversation history and primary sources (code or specifications) as specified in SKILL.md.
Boundary markers: No specific delimiters or instructions to ignore embedded content are used when interpolating ingested data.
Capability inventory: The agent performs shell command execution using the vp dlx or pnpm dlx command on the generated file.
Sanitization: No sanitization or validation mechanisms are described for the external data before it is processed.

teach-lesson