The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes several scripts (scripts/bootstrap.ts, scripts/orch/orch.ts, scripts/watch-pr/watch-pr) that execute shell commands such as git, gh (GitHub CLI), and bun install. These are used for standard repository management, pull request monitoring, and dependency bootstrapping within the skill's own directory. The logic includes safeguards like --frozen-lockfile and specific path resolution.
[METADATA_POISONING]: The SKILL.md file contains instructions aimed at the AI agent to prioritize its own rules ('Start every multi-step task with a todolist whose first item is to read the Principles section...'). While these are structural instructions for agent behavior, they are typical for 'mode' or 'persona' skills and do not attempt to bypass platform-level safety filters.

rstack-mode