The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes automated shell commands for project management and development workflows.
Evidence: scripts/missions.sh contains logic to detect and run project scripts like npm run test, lint, and typecheck based on package.json content.
Evidence: The script manages Git worktrees using git worktree add, remove, and merge commands.
[REMOTE_CODE_EXECUTION]: Sub-agents are launched with high-privilege flags that bypass standard security boundaries.
Evidence: scripts/missions.sh executes the claude CLI with the --dangerously-skip-permissions flag.
Evidence: scripts/missions.sh executes the codex CLI with the --dangerously-bypass-approvals-and-sandbox flag.
Technical Impact: This configuration allows sub-agents to perform any tool-based action (file modification, network requests, command execution) without requesting user permission, enabling autonomous operation at the cost of significant security oversight.
[EXTERNAL_DOWNLOADS]: The skill automates the installation of external software dependencies.
Evidence: The pm_install function in scripts/missions.sh runs package manager install commands (npm install, pnpm install, etc.) autonomously when setting up worker worktrees.

missions