The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to automatically execute a shell command (ls ~/.claude/skills/ ~/.agent/skills/ ...) upon its first activation in a conversation to verify the presence of recommended companion skills.
[EXTERNAL_DOWNLOADS]: The README and SKILL.md files promote the installation of additional remote code packages from the 'AbsolutelySkilled' repository using the npx skills add command.
[DATA_EXFILTRATION]: The skill provides detailed guidance and command examples for network-capable utilities like curl, tcpdump, and strace which, while legitimate for debugging, provide the capability to capture or transmit sensitive system data.
[PROMPT_INJECTION]: The 'Companion check' routine introduces a surface for indirect prompt injection by instructing the agent to process the output of local shell commands (filenames) as instructions for its next action without using sanitization or boundary markers.
Ingestion points: File names from the output of the ls command executed in the skill activation sequence.
Boundary markers: Absent.
Capability inventory: The skill has access to shell execution (ls), system tracing (strace), network monitoring (tcpdump), and process control (kill).
Sanitization: Absent.

debugging-tools