交互式引导配置多 Agent 系统，支持批量创建、凭证验证、角色模板、自动备份

This skill mostly does what it says (creates and configures Feishu bots and updates OpenClaw config), but there are a few red flags you should address before installing or running it: - Do NOT paste AppSecret (or any secret) into open chat without understanding where it will be stored. The skill instructs users to reply with App ID/App Secret in the chat, and the code writes those credentials into ~/.openclaw/openclaw.json and creates backups — these files can be read by processes or users with access to that account. - Manually inspect SKILL.md (raw) for hidden characters or base64 blobs flagged by the pre-scan. Remove any unexpected control/unicode characters or opaque blocks before use. - Review the bundled code (dist/index.js or src/index.ts) yourself to confirm it only performs the described file operations (backups, config updates, directory creation) and does not exfiltrate data or call unexpected endpoints. The code provided appears to be local FS operations and validation, but you should verify the full file contents (including any truncated sections) and run tests in a safe environment. - Make a separate backup of your existing ~/.openclaw/openclaw.json before running the skill. The skill creates its own backups, but an independent copy ensures safety. - Consider running this first in a sandbox/test environment or container and not on a production host. Check resulting config files for secrets and permissions and rotate any AppSecrets used for testing. - If you are uncomfortable having the agent modify config or store secrets in plaintext, ask the author to support secure secret injection (environment variables, secret store, or ephemeral input that isn't logged) or avoid installing. If you want, I can: (1) extract and show the raw SKILL.md so you can inspect for hidden characters, (2) list exactly where in the code credentials are written, or (3) suggest safer workflows for providing secrets (e.g., secrets API or environment variables).

Type: OpenClaw Skill Name: feishu-multi-agent-manager Version: 2.0.5 The skill is an interactive configuration assistant designed to help users set up multiple Feishu (Lark) agents within the OpenClaw environment. It performs high-privilege operations such as reading and modifying the main 'openclaw.json' configuration file and creating workspace directories in the user's home directory. However, these actions are strictly aligned with the stated purpose of the skill. The code includes safety measures like automatic configuration backups and credential format validation. No evidence of data exfiltration, unauthorized remote execution, persistence mechanisms, or malicious prompt injection was found in files like 'src/index.ts' or 'SKILL.md'.