CommentAI - Typecho Blog AI Intelligent Comment Moderation & Reply Plugin- Zeruns's Blog

CommentAI - Typecho Blog AI Intelligent Comment Moderation & Reply Plugin

Introduction

CommentAI is an AI intelligent comment moderation & reply plugin for Typecho blogs.

Core Features

Automatically use AI to moderate user comments, move rejected comments to pending review for manual approval
Automatically generate comment replies
Three working modes available (auto-reply, manual review, suggestion only)
Support for Qwen, OpenAI, DeepSeek, Kimi and other model settings
Support for context-aware replies
Support for sensitive word filtering and frequency limiting
Support for reply delay
Support for trigger condition settings
Support for AI identifier display

Based on the CommentAI plugin developed by blogger xuan, with secondary development using AI, adding comment AI moderation functionality to the original plugin (which only had AI reply functionality).

Original plugin developer's blog post: https://blog.ybyq.wang/archives/1527.html

Open source link for the secondary developed plugin: https://github.com/zeruns/CommentAI

Installation

Environment Requirements

Typecho 1.2.0 - 1.3.0
PHP 7.0+ (requires curl, json, mbstring extensions)
MySQL 5.7+ or SQLite 3.0+
AI service API Key

Installation Steps

Upload the CommentAI folder to the /usr/plugins/ directory
Backend "Console" → "Plugins" → Enable "CommentAI"
The plugin automatically creates database tables
If creation fails, manually execute the install.sql file

Configuration

Basic Configuration

Plugin Switch

Enable/disable the plugin

Reply Mode

Fully automatic: AI generates and directly publishes
Manual review: requires manual approval after generation
Suggestion only: for reference only, not published

Administrator UID

User identity used for AI replies, default is 1

AI Platform Configuration

Supported Platforms

Alibaba Cloud Bailian: fast domestic access, affordable price
OpenAI: high quality, requires international network
DeepSeek: high cost-performance ratio
Kimi (Moonshot): domestic large model
SiliconFlow: one-stop cloud service platform integrating top large models
DMXAPI: Chinese multimodal large model API aggregation platform
Custom: support for OpenAI compatible interfaces

API Configuration

API Key: required, obtain from the corresponding platform
API Address: optional, leave blank to use default
Model Name: such as qwen-plus, gpt-4o-mini, deepseek-chat, moonshot-v1-8k

Prompt Configuration

System Prompt

Define the AI's role and reply style:

You are a friendly, professional blogger. Generate appropriate replies based on reader comments.

Requirements:

Natural, friendly tone
Provide valuable responses to comment content
Give clear answers to questions
Reply length 50-150 words
Reply in Chinese

Context Information

Include article title
Include article summary (increases token consumption)
Include parent comments

Advanced Configuration

Temperature Parameter

0.0-0.3: strong determinism
0.4-0.7: balanced (recommended)
0.8-1.0: strong creativity

Maximum Token Count

Recommended 200-500

Sensitive Word Filtering

One sensitive word per line

Frequency Limit

Maximum number of calls per hour, 0 for no limit

Reply Delay

How many seconds to delay reply, 0 for immediate reply (recommended)
Recommended 30-120 seconds

Display Settings

AI Identifier

Display: add identifier at the end of reply (recommended, customizable)
Do not display: looks like manual reply

Trigger Conditions

Only reply to approved comments (recommended)
Ignore spam comments (recommended)
Ignore quotes and Trackback (recommended)
Only reply to the first comment of the article
Automatically exclude administrator comments

AI Moderation Configuration

API service provider can be set separately, you can choose a smaller model to save money, or some free models

Usage

Workflow

Reader posts a comment
Plugin checks trigger conditions
Triggers AI moderation
Delays for specified time (if configured)
Calls AI to generate reply
Processes according to mode

Management Panel

Access: Backend → AI Comment Reply

Features

Statistics cards: display counts for each status
Status filtering: view replies in different statuses
Queue list: view detailed information
Operation buttons: publish, reject, regenerate, view original comment

Toolbar

Refresh: reload data
Clean up old records: delete records older than 30 days
Plugin settings: jump to configuration page
Test connection: test AI service

Comment AI moderation effect:

Uninstallation

Disable the plugin in the backend
Delete the plugin folder
(Optional) Delete database table:

DROP TABLE IF EXISTS `typecho_comment_ai_queue`;

Download Links

Github: https://github.com/zeruns/CommentAI

123 Cloud Disk: https://www.123865.com/s/2Y9Djv-5FddH?pwd=ffLm#

目录CONTENT

CommentAI - Typecho Blog AI Intelligent Comment Moderation & Reply Plugin