The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides scripts and instructions for executing local processes via tmux sessions and Python's subprocess and pty modules.
Evidence:
Bash scripts using tmux new-session and tmux send-keys to drive an application.
A Python PTY harness template utilizing subprocess.Popen with pseudo-terminals to programmatically interact with a command-line tool.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and act upon output captured from the terminal of the command under test, which presents a surface for indirect prompt injection if the tested tool emits malicious patterns.
Ingestion points: Terminal output is captured via tmux capture-pane and os.read(master_fd) in SKILL.md.
Boundary markers: None present in the templates provided.
Capability inventory: The skill allows for file writing (in /tmp), network inspection (local Node profiling), and command execution.
Sanitization: No explicit sanitization or filtering of terminal output is documented in the harness templates.

control-cli