developer-community-moderation

Start from an established base document like Contributor Covenant 3.0 or the Django code of conduct rather than writing from scratch. Rewrite the enforcement section to describe the procedure your team will actually follow, name real reporting channels, and publish a ladder with a trigger and consequence per rung.

Acknowledge the report within a published window, preserve evidence before anything is deleted, apply recusal rules, then interview the reporter, witnesses, and the reported person in that order. Decide with a rule fixed in advance and notify with a four-part message: behaviour, impact, expected change, consequence.

Prefer 3.0 for a new document because it drops 2.1's line suggesting a public apology may be requested, which enforcement guidance warns against. The two versions use different section names, so state which one you are adapting when you copy text.

Gate volume rather than trying to detect authorship, since there is no reliable technical indicator for AI-generated content. Rank controls by inflow removed per moderator hour: rate and interaction limits first, then entry gates and reputation requirements, with individual bans last.

Do not ship it. Practitioner guidance holds that a code must apply to the community's most powerful people, and a selectively enforced code is worse than none. The honest options are removing the exemption or publishing community guidelines without enforcement language.

No. Publish a reporter-facing summary of what to expect, in what order, and by when, but keep the operational playbook internal and revisable. A published step-by-step procedure gets used to argue the team failed to follow its own process, which stalls enforcement.