The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill instructions include executing cargo test and cargo run. This executes arbitrary code within the workspace context, which could lead to code execution vulnerabilities if the repository contains malicious scripts or dependencies.
[COMMAND_EXECUTION] (HIGH): The skill utilizes tshark -i any for background packet capture. Capturing on 'any' interface typically requires root/sudo privileges and allows the agent to monitor all unencrypted network traffic on the host, including sensitive data from unrelated processes.
[EXTERNAL_DOWNLOADS] (SAFE): The prerequisites recommend installing tshark via apt, which is a trusted system package manager.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill processes untrusted external data in the form of network packets.
Ingestion points: Raw hex data is read from hses_export_{timestamp}.txt for protocol analysis.
Boundary markers: Absent. No delimiters or instructions are used to prevent the agent from interpreting packet data as commands.
Capability inventory: File system access (mkdir), network monitoring (tshark), process management (pkill), and code execution (cargo).
Sanitization: Absent. The skill processes hex data directly into a markdown report without validation or escaping.

hses-packet-analysis