The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a directory listing command (ls) on specific paths such as ~/.claude/skills/ to identify which recommended companion skills are currently installed. This is a benign environmental check used to provide helpful tool recommendations to the user.\n- [PROMPT_INJECTION]: The skill demonstrates an attack surface for indirect prompt injection as it is designed to ingest and analyze untrusted external data like SDK code and documentation. Ingestion points: User-provided SDK source code and documentation text (SKILL.md). Boundary markers: Absent; the instructions do not specify the use of delimiters or warnings to ignore instructions embedded in the processed user content. Capability inventory: The agent is instructed to perform filesystem discovery (ls) and suggest tool installation commands (npx) to the user. Sanitization: Absent; no methods for escaping or validating external content are described in the skill instructions.

developer-experience