The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to run language-specific development utilities such as LSPs (go vet, mypy, tsc), linters (eslint, ruff, golangci-lint), and test runners (pytest, vitest, cargo test). These are standard operations for an AI development agent.\n- [EXTERNAL_DOWNLOADS]: The workflow includes steps to install necessary development dependencies and tools using standard package managers (npm, pip, npx). It specifically references agent-browser from vercel-labs, which is a trusted organization, and playwright, a well-known testing framework.\n- [REMOTE_CODE_EXECUTION]: The skill utilizes the platform's Agent and Agent Teams capabilities to delegate complex tasks to specialized sub-agents. This is the intended architecture of the platform and is used here to separate planning, implementation, and quality verification roles.\n- [DATA_EXFILTRATION]: While the skill can route requests to a 3rd party API (Z.AI) for cost optimization in 'CG Mode', this is a documented feature that requires explicit user setup (providing an API key) and environment configuration. Standard network operations are limited to well-known domains like GitHub via the official gh CLI.\n- [COMMAND_EXECUTION]: The skill uses the '!' syntax in SKILL.md to run benign git commands (git status, git branch) to provide the agent with project context upon loading. These commands are static and do not accept user-supplied arguments.

moai