Roundtable Skill is a workflow skill, docs, templates, and install helpers.

It does not bundle Lingtai, run a hosted service, collect telemetry, store credentials, or require users to paste secrets into this repository.

Open a GitHub issue for public security concerns that do not include secrets or private runtime data.

Do not paste:

• API keys
• OAuth tokens
• private keys
• cookies
• database URLs
• mailbox contents
• .lingtai/ runtime files
• codex-auth.json
• user data
• private project paths

If a report requires private material, open a minimal issue asking for a secure contact path first.

• .lingtai/
• .recipe/
• mailbox files
• agent logs
• OAuth tokens
• auth JSON files
• .env
• private keys
• runtime data

The Codex install scripts copy the Roundtable skill files into the local Codex skills directory. They do not install Lingtai, create agents, request secrets, or configure model/provider credentials.

Lingtai setup is explicit and external:

• Lingtai repo: https://github.com/LingTai-AI/lingtai
• Setup guide: docs/LINGTAI_SETUP.md

• accidental secret leakage in examples or docs
• unsafe install instructions
• committing runtime agent state
• misleading claims about automatic multi-agent execution without Lingtai
• scripts that write outside the intended local skill directory

Before publishing a release:

• run the Docs GitHub Action,
• run scripts/check-roundtable.*,
• follow docs/MAINTAINER_RELEASE.md,
• inspect tracked files,
• confirm no .lingtai/, .recipe/, mailbox, auth, env, key, or log files are tracked,
• confirm examples use placeholders only,
• confirm README wording does not claim Roundtable works without Lingtai.