The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install the @oh-my-pi/pi-team package globally via npm. This package is not associated with a verified trusted organization or well-known service provided in the analysis context.
[COMMAND_EXECUTION]: The skill relies on the pi-team CLI to execute system-level commands, manage tmux sessions, and interact with the file system (e.g., creating the .pi-team/ directory).
[COMMAND_EXECUTION]: Troubleshooting instructions include the use of sudo apt install tmux, which requests elevated administrative privileges to install system prerequisites.
[PROMPT_INJECTION]: The orchestration of multiple AI providers (PI, Claude, Codex, Gemini) through a single aggregation point creates a surface for indirect prompt injection.
Ingestion points: Output logs from individual agents in .pi-team/worker-*/output.log are ingested by the pi-team aggregate command.
Boundary markers: The skill documentation does not specify the use of delimiters or instructions to ignore embedded commands during the aggregation process.
Capability inventory: The skill possesses the ability to execute shell commands via tmux and perform file writes, which could be abused if an agent produces malicious instructions.
Sanitization: No evidence of input validation or output sanitization is provided for the data collected from remote AI providers.

pi-team