- Home
- Minecraft Moderation Plugins
- AdvancedSensitiveWords
AdvancedSensitiveWords
What it is
AdvancedSensitiveWords is a free, open-source moderation plugin for self-hosted Minecraft servers running Paper. It uses DFA word matching and event-based inspection to check chat, commands, writable books, signs, anvil results, item names and lore, player names, and optionally broadcasts. Operators can replace or cancel matched content and configure responses such as commands, damage, effects, hostile behavior, or shadowbans. Version 2.x targets Paper 1.21.11 and requires Java 21. It can be installed from a release artifact or built locally and placed in the server's plugins directory. Optional LLM review is off by default and checks messages that pass DFA and chat-context checks; it requires a compatible provider and API key. The review process includes message-length, entropy, cooldown, concurrency, and queue limits. Request and response records are kept in a local directory that should be treated as sensitive operational data. The optional Velocity module supports notifications and proxy commands, but does not filter chat on its own. The project is licensed under GNU AGPL-3.0.
For and against
For
- Checks chat and multiple in-game text surfaces.
- Matched content can be replaced or canceled.
- Offers configurable actions, including shadowbans.
- Optional LLM review has operational safeguards.
- Other Paper plugins can use its shadowban API.
Against
- Version 2.x requires Paper 1.21.11 and Java 21.
- Spigot, CraftBukkit, and BungeeCord are unsupported.
- LLM review requires a provider and API key.
- LLM audit records are sensitive operational data.
The verdict
AdvancedSensitiveWords provides Paper server operators with broad content inspection and configurable responses. Its Paper-only compatibility and the handling required for optional LLM audits are important setup considerations.
Who it suits
It suits Minecraft server operators using Paper who need configurable content checks and moderation actions. Operators considering AI review should also be prepared to configure a compatible provider and handle its locally stored audit data.
On the record
- AI privacy handling
- LLM requests and responses are audited in a local llm-history directory that the README says to treat as sensitive operational data.github.com · read 4 Oct 2026
- AI review
- Optional LLM chat review is disabled by default and can review messages that pass the DFA and chat-context checks.github.com · read 4 Oct 2026
- AI safeguards
- LLM review applies message-length, entropy, cooldown, concurrency, and queue limits, and locally parses model output before follow-up actions.github.com · read 4 Oct 2026
- API
- Other Paper plugins can use its shadowban API, and it fires an asynchronous moderation response event after an LLM response.github.com · read 4 Oct 2026
- Compatibility limit
- Since version 2.x, Spigot, CraftBukkit, and BungeeCord are unsupported.github.com · read 4 Oct 2026
- Content filters
- It can check chat, commands, books, signs, anvils, item names and lore, player names, and broadcasts.github.com · read 4 Oct 2026
- Distribution
- The README instructs users to download the Paper artifact from a release or build paper:shadowJar locally.github.com · read 4 Oct 2026
- Installation
- The Paper artifact can be downloaded from a release or built locally, then placed in the server plugins directory.github.com · read 4 Oct 2026
- Integrations
- The README lists compatibility with TrChat, optional PacketEvents, PlaceholderAPI, Floodgate, and AuthMe.github.com · read 4 Oct 2026
- Intended users
- Installation and configuration instructions describe placing the plugin on a Paper server and setting its server-side configuration, indicating it is for Minecraft server operators.github.com · read 4 Oct 2026
- License
- The project is licensed under GNU AGPL-3.0.github.com · read 4 Oct 2026
- LLM providers
- The configuration supports CHAT_COMPLETIONS, RESPONSES, or ANTHROPIC_MESSAGES API modes and gives DeepSeek as an example base URL.github.com · read 4 Oct 2026
- Maker identity
- The repository is published under GitHub account HaHaWTH, whose profile displays the name Creeam HaHaWTH.github.com · read 4 Oct 2026
- Moderation actions
- Matched content can be replaced or cancelled, and configured violation levels can trigger actions such as commands, damage, effects, or shadowbans.github.com · read 4 Oct 2026
- Operational data
- LLM requests and responses are audited in the plugin's llm-history directory, which the README says to treat as sensitive operational data.github.com · read 4 Oct 2026
- Optional AI review
- LLM chat review is disabled by default, uses a compatible provider and API key, and only reviews messages missed by DFA and chat-context checks.github.com · read 4 Oct 2026
- Platform limit
- Version 2.x targets Paper only; the README says Spigot, CraftBukkit, and BungeeCord are unsupported.github.com · read 4 Oct 2026
- Proxy limitation
- The Velocity module supports notifications and proxy commands but does not filter chat on its own.github.com · read 4 Oct 2026
