feishu-multi-agent

This is a practical blueprint for running multiple Feishu-backed agents and appears to be what it says, but pay attention to a few things before following it in production: - Metadata mismatch: the registry lists no required credentials, but the guide requires creating multiple Feishu apps and storing appId/appSecret values in your OpenClaw config. Treat those as sensitive secrets and do not commit them to source control. - Platform/tools assumptions: the guide uses macOS launchctl and references an `agent-browser --cdp-endpoint` automation technique. Verify those commands/tools apply to your environment or adapt them safely. - Principle of least privilege: when creating Feishu apps, grant only the permissions each bot needs. Avoid giving broad admin rights to bots that will be scheduled autonomously. - Workspace file access: the approach relies on using bash to read/write across agent workspaces. That requires filesystem access and can be powerful — test in an isolated environment first and confirm access boundaries are intentional. - Cron/autonomy risk: the recipe creates recurring autonomous jobs for agents (cron jobs + system-events). Audit jobs and sessions after creation to avoid runaway or misbehaving agents. If you plan to install/use this blueprint, ask the author to clarify the exact credentials/files you must provide, list platform/tool dependencies explicitly, and document secure handling of app secrets. Running a staged test on an isolated machine is recommended.