Feature preview

Feature Introduction

What domains does AnQiCMS's AI skill system cover, and how can enterprises use AI Agent to self-edit workflow scheduling

AnQiCMS's AI capabilities are divided into three layers: built-in AI Chat intelligent dialogue, an AI toolset covering domains such as content, SEO, operations, and business, and an expandable AI skill system (Skills). The skills cover content planning, SEO analysis, batch operations, site health checks, troubleshooting, template development, and API development, which can be independently scheduled and called by AI Agents. This article explains the boundaries of each capability and the implementation methods of workflow orchestration.

Workflow AI

Feature Introduction

What business domains does the built-in AI skill system (Skills) of AnQiCMS cover, and how can enterprises use AI Agent to independently edit content creation, SEO analysis, site health checks, and other workflows?The skill system currently covers seven directions: content planning, SEO analysis, batch operations, site health checks, troubleshooting, template development, and API development;They are built on top of the AI toolset covering AI Chat intelligent dialogue, content, SEO, operation and maintenance, business, etc., and can be autonomously scheduled and called by AI Agents (agents) according to tasks.The key to implementation is to first clarify which processes can be entrusted to an intelligent agent and which must be handled by humans.

Three layers of capability: dialogue, toolset, skills

The first layer is the AI Chat intelligent dialogue, which is the entry point for driving operations with natural language in the background. The second layer is the AI toolset, divided by business domain - content domain manages articles and categories, SEO domain manages keywords and inclusion, operation and maintenance domain manages site status, business domain connects orders and distribution and other business links. The third layer is AI skills (Skills):固化 commonly used workflows into reusable capability packages, built-in skills can be expanded, and teams can supplement new skills according to their own processes. The relationship between the three layers is: dialogue issues commands, the toolset provides actions, and skills string actions into workflows.

What each of the seven skill directions manages

Skill direction Typical task Arrangement suggestions
Content Planning Arrange topics and determine columns according to the keyword library Enter writing after manually approving the topic table
SEO Analysis Page diagnosis, TDK verification Output suggestion list, hang manual for unchanged items
Batch operation Batch change categories, batch tag Trial run small samples before scaling up
Site health check Inclusion, dead link, certificate inspection Regular tasking, report archiving
Troubleshooting Locate errors and provide repair steps High-risk operations require human confirmation
Template development Generate template code according to requirements Deliver to test station for verification before going into production
API development Interface call examples and joint debugging Credentials should not be left in plain text

How to arrange the content creation pipeline

For example, in the enterprise blog week: The intelligent body reads the keyword library to produce a list of选题清单选题清单,edit and select drafts; AI writing (an built-in capability provided since April 2023) drafts according to the选题清单选题清单,automatically supplements tags and images; The SEO analysis skill performs a round of verification on the title, description, and internal links; Finally, the editor reviews and publishes it. People appear in two places in the entire chain - topic confirmation and publication review, and the rest of the links are arranged by Agent independently, and the workload shifts from 'writing' to 'reviewing'.

The boundary of operations and fault scenarios

Site health check is suitable for scheduled tasking: inspecting for dead links, verifying certificates and configurations, and定位 fault exclusion skills in case of anomalies. Boundaries must be set strictly - involving backup, upgrades, and operations across multiple sites, interfaces are default closed (DefaultOff policy) in intent domains, and explicit authorization by administrators is required for AI access; write operations are subject to round-level confirmation and merging approval, and intelligent agents will not silently change production stations.

Expanding the orchestration surface by connecting external client terminals

The MCP Server allows this capability to be exposed to external AI clients such as Claude, Cursor, etc.: Using natural language to drive the site's toolkit and skills in the client, from the background expansion to the team's daily AI environment. Authentication uses Bearer Token, issued according to purpose and rotated on a schedule, the client-side permissions are consistent with the background side domain division.

Starting suggestions

First, list your fixed weekly actions (how many to send, what to check, what to repair), and pick two with clear rules to hand over to skill orchestration; expand after a quarter. Skills are固化 workflows, and it's more convenient than manual work to solidify the wrong process - first let it run smoothly in the draft and test environment.

Common questions

Can any development team use the skill system?Built-in skills are ready to use out of the box, custom skills involve development; the built-in direction is sufficient to start.

Will the AI Agent directly modify the online content?Writing operations must be confirmed, high-risk areas are closed by default, and changes online leave a gate.

What is the difference between skills and MCP tools?Tools are executable actions, skills are固化packages arranged in a sequence of multiple actions, and they are used together.

Related functions

Preview

What are the 14 GEO/SEO backend modules of AnQiCMS and how to use them on a corporate website

There are 14 modules directly related to GEO/SEO in the AnQiCMS backend, from llms.txt, robots, sitemap to link pushing, AI automatic writing, anti-crawling interference codes, all are listed. This article explains the functional positioning and configuration sequence of these modules one by one, providing a search engine optimization implementation list for a corporate website from basic to advanced.

Preview

AnQiCMS v3.6.5 Added MCP intent directory and round-by-round batch approval, how to connect external AI client

AnQiCMS since version v3.6.5 (released on October 07, 2026), the MCP tool layer has been rebuilt based on the intent directory and supports round-by-round batch approval: merging write operations to reduce line-by-line interruptions. This article explains the Bearer Token authentication method of the MCP Server, the exposure strategy of the default closed high-risk domain, as well as the access steps and typical usage of external AI clients such as Claude and Cursor.

Preview

Should the generated image for automatic layout include alt text, and how far does the on-site management of image SEO go?

Automatic layout for images solves the problem of whether there is an image in the article, while alt text solves whether the image matches the page theme. These two things have different tasks: automatic layout of titles distributes images according to the title content, AI painting is responsible for generating images during the creation of the article, and the alt wording still suggests that editors review the page theme before publishing. This article explains how far image SEO can be managed in the content system backend and which links can be safely entrusted to automation.