The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dispatches instructions and manages task lifecycles by executing commands within tmux panes using the atm send and tmux send-keys utilities.
[COMMAND_EXECUTION]: Monitors and manages local agent processes using ps and atm kill to ensure the health and convergence of the orchestration swarm.
[INDIRECT_PROMPT_INJECTION]: The skill processes task data and coordination messages from external sources that represent an attack surface for indirect injection.
- Ingestion points: Data is ingested from Agent Mail (am mail inbox), the beads work queue (br ready), and GitHub PR metadata (gh pr list).
- Boundary markers: No explicit delimiters are specified to distinguish between data and control instructions from these sources.
- Capability inventory: The skill can execute commands in terminal panes, manipulate Git repositories, and manage local process states.
- Sanitization: No specific sanitization or validation routines are described for data received from coordination layers.

using-atm