Release Date: 11/2025
Version: V4.1
Update Summary
This V4.1 release focuses on comprehensive upgrades in AI Studio (Agent module), system and security management, and enhanced orchestration capabilities.
Multiple new features for Agent and token management have been added, document translation and security policies have been optimized, and the system's flexible configuration and security control system have been further improved, comprehensively enhancing the platform's intelligence and manageability.
Main Updates
AI Studio / Agent Management
Core Highlights
Supports multi-level Agent and MCP management from individual to organization level, provides pre-built templates, intelligent document translation capabilities (including PDF), App and Skill management entry, and allows Agents to be published as MCP Servers callable by other Agents, enhancing orchestration and reusability.
Latest Features
-
Multi-level Agent/MCP Creation: Regular users can create personal Agents/MCPs, while admins/super admins can create advanced, shared, and organization-level Agents/MCPs (supporting advanced orchestration with UI).
-
Configurable Bottom Menu: Supports configuring custom bottom menus and jump addresses in the backend via environment variables.
-
Pre-built Agent and MCP Templates: Provides industry-categorized pre-built Agent/MCP templates to accelerate creation and deployment.
-
AI Translator (Document Translation): Supports fast translation of PDF documents, importing local corpora/language packs, and intelligent translation and alignment of document content by language.
-
App Management Optimization: Added Deep Research entry; supports PDF translation download, corpus import, and knowledge preview; retains access to existing apps via Link address (frontend direct creation of new Apps is not yet supported).
-
Skill and Model Marketplace Pre-built: Opens up small language models and machine learning model capabilities (Coming Soon), facilitating the expansion of various model types.
Identity & Security
Core Highlights
Establishes a unified and controllable permission and token system, enhances personal privacy (PII) identification and content review capabilities, provides centralized management of sensitive words and content review models, and improves the platform's security and compliance capabilities.
Latest Features
-
PII Personal Identity Information Recognition and Policy: Supports identification and configuration of prompt, masking, and prohibition policies for sensitive personal information.
-
Unified Token Management (Access Token): Upgrades decentralized API Key management to a centralized Access Token system, supporting binding to users/permissions and lifecycle management.
-
Token Support Scope: Access Token API supports: Agent configuration queries, FAQ list OpenAPI, and MCP Server calls published by Agents.
-
Content Review Model and Sensitive Word Management: Supports enabling content review mode, setting confidence thresholds, and adding, deleting, modifying, and querying sensitive words for platform-level control.
System & Monitoring
Core Highlights
Improves environment variable, license, and system prompt management capabilities, strengthens logging and monitoring, and supports recording key operation logs by product function for auditing and troubleshooting.
Latest Features
-
License Management: Clearly displays user, number of users, number of enterprise Agents, and validity period to avoid overuse.
-
System Prompt Maintenance: Displays and maintains all configurable system prompts across the platform in a list format.
-
Enhanced Log Management: Records core module logs by product function (including knowledge base, data management, permission management, sensitive words, Agent calls, etc.), facilitating query and audit.
Basic & Advanced Orchestration and MCP Service
Core Highlights
Enhances basic and advanced orchestration capabilities, supports node-level system parameter input, and allows Agents to be published as MCP Servers for invocation by other Agents or external systems, improving reusability and service capabilities.
Latest Features
-
MCP Server Publishing Capability: Supports publishing Agents as MCP Servers, serving as service nodes callable by other Agents (V4.1).
-
Basic Orchestration Enhancement: Supports configuration of whether to output file content and preview/debug of set content.
-
New Advanced Orchestration Features: Adds configuration nodes to support system parameter input, supports editing nodes for system parameter input and debugging tools, and supports trial runs and error detection of running flows.
-
PWA / Entry Optimization: Supports packaging URLs as desktop icons for easier access.