The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions mandate that the agent use npx to download and execute the modern-web-guidance package from the NPM registry at runtime. This involves running code from an external repository directly within the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes and adopts natural language instructions from externally retrieved guides. This creates a surface where content from the remote package could influence the agent's subsequent code generation or project decisions.
Ingestion points: SKILL.md instructs the agent to run retrieve commands whose output is guide content from the remote package.
Boundary markers: Absent; the guide content is treated as authoritative and is not wrapped in markers to distinguish it from the agent's instructions.
Capability inventory: The agent can generate code, modify files, and execute shell commands based on retrieved content.
Sanitization: Absent; the skill does not verify the content of the retrieved guides before processing them.
[PROMPT_INJECTION]: The skill's frontmatter and instructions use directive markers such as 'MANDATORY: Execute FIRST' and 'Do NOT skip' to force the agent to prioritize the tool over its own internal weights and safety guidelines.

modern-web-guidance