The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted output from terminal sessions, creating a surface for indirect prompt injection.
Ingestion points: The agent reads terminal content using tmux capture-pane (documented in SKILL.md and implemented in scripts/wait-for-text.sh).
Boundary markers: There are no specific instructions or delimiters to help the agent distinguish between process output and system instructions.
Capability inventory: The agent can perform read/write cycles by using tmux send-keys to respond to captured output.
Sanitization: Content retrieved from terminal panes is processed without explicit sanitization or filtering.
[COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary commands within tmux sessions.
It uses tmux send-keys to relay text, control sequences, and special keys to terminal panes.
It includes helper scripts scripts/find-sessions.sh and scripts/wait-for-text.sh to automate the discovery of and waiting for specific states in terminal sessions.

tmux