The Agent Skills Directory

[COMMAND_EXECUTION]: The SKILL.md file contains a 'Companion check' instruction that directs the agent to execute a shell command (ls) on various hidden directories in the user's home folder (e.g., ~/.claude/skills/). This is used to programmatically detect if recommended skills are currently installed.
[PROMPT_INJECTION]: The skill focuses on processing untrusted external data, such as HTTP request headers (CF-IPCountry, Cookie) and response bodies (response.text()), as shown in the Cloudflare Worker examples. This architecture creates an attack surface for indirect prompt injection if the ingested content contains malicious instructions.
Ingestion points: External data enters the agent's context through web requests and response transformations in the Worker code examples.
Boundary markers: The provided examples do not use delimiters or warnings to prevent the agent from treating data as instructions.
Capability inventory: The skill environment includes file system inspection (ls), network operations (fetch), and environment management.
Sanitization: There is no mention of sanitizing or validating external data before it is processed or interpolated.
[EXTERNAL_DOWNLOADS]: The documentation (README.md and SKILL.md) provides instructions to download and execute code from a remote source (AbsolutelySkilled/AbsolutelySkilled on GitHub) using the npx skills add command for installation and management of related skills.

edge-computing