The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command (ls) to probe the file system for specific configuration directories (e.g., ~/.claude/skills/, ~/.agent/skills/). This behavior constitutes environment discovery, revealing information about the user's installed tools and configuration.
[REMOTE_CODE_EXECUTION]: The skill proactively suggests the use of npx to install additional remote packages from the AbsolutelySkilled repository. While the vendor matches the skill author, the automated check of the user's environment followed by a prompt to execute remote code increases the risk of unauthorized software installation.
[PROMPT_INJECTION]: The skill is designed to ingest and process external project data, such as risk registers and status updates, creating a surface for indirect prompt injection.
Ingestion points: Processes project plans, risk descriptions, and RAID logs provided by the user (described in SKILL.md).
Boundary markers: Absent; there are no instructions to delimit or ignore instructions within the processed project data.
Capability inventory: The skill utilizes shell command execution via the ls command in the Companion check section of SKILL.md.
Sanitization: No evidence of input validation or sanitization is present for the data being processed.

project-execution