@ykaixu

openclaw-skill-reviewer audits skills across format, content quality, and functional correctness via a four-level process combining automated validation, manual review, template verification, and best-practices enforcement.

rssh2 wraps ssh2 for session pooling, tunnel management, and sftp operations. covers basic command execution, port forwarding, and file transfer but lacks structured intent mapping and explicit error recovery workflows.

agent-builder-plus scaffolds openclaw agent workspaces via interview and file generation. covers identity, soul, agents, user, heartbeat, and optional memory files. includes substantial operational guidance on channel binding, model configuration, and deployment, but procedure lacks clarity on when to use the skill versus how it actually executes.

skill-reviewer-pro documents a six-level review framework for openclaw skills but fails to implement its own stated standards. the skill itself lacks operational procedures, concrete trigger phrases, and actionable edge case handling.

安全地管理 OpenClaw 模型配置。用于添加、测试和配置新模型到 models.json，包括 API key 验证、模型可访问性测试、工具调用功能检测、设置默认模型和配置到特定 agent。所有操作都会自动备份配置文件以确保安全。