The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/relay.mjs file utilizes node:child_process to execute external binaries including the omp implementer, git for file tracking, and taskkill on Windows. It also executes a script from a sibling skill (../../delegate-setup/scripts/lane.mjs) to resolve fleet configurations, introducing a dependency on external skill logic.- [REMOTE_CODE_EXECUTION]: The skill is designed to drive the @oh-my-pi/pi-coding-agent, which is an autonomous agent capable of executing shell commands and modifying the filesystem. The relay script specifically passes the --yolo flag, which allows the agent to execute its tools without user confirmation. There is no sandbox provided for this execution environment.- [EXTERNAL_DOWNLOADS]: The instructions require the user to install a third-party implementer tool from https://omp.sh or via bun install -g @oh-my-pi/pi-coding-agent. These sources are outside the verified trusted vendor list.- [INDIRECT_PROMPT_INJECTION]: The implementer tool is configured to automatically ingest context from files within the workspace, such as AGENTS.md and CLAUDE.md. These files represent an attack surface where malicious instructions could be embedded to influence the agent's behavior during a delegated task. The skill relies on a 'review-and-land' workflow as the primary mitigation for this risk.

omp-delegate