2026

Feature

Description

References

Long-term Memory Service

The Long-term Memory Service is a core capability provided by the Context Service. It manages memory data and service configurations based on ADBPGMem, and supports three memory scenarios: Session, User, and Agent.

Long-term memory service

Long-term Memory Integration with Coding Agent

The AnalyticDB for PostgreSQL Long-term Memory Skills are a set of development knowledge management tools that run within a Coding Agent. Each Skill is a capability module in an AI IDE that can be triggered via slash commands (such as /adbpgmem-record) or natural language (such as "record this"). By connecting to the AnalyticDB for PostgreSQL Long-term Memory Service API, it enables AI assistants to capture, retrieve, and maintain tacit knowledge generated during daily development conversations — knowledge that is not in the code but is critical to the project, such as experience, decisions, and lessons learned.

• Integrate Long-Term Memory Skills with Coding Agent

• Integrate Long-Term Memory Skills with Coding Agent

• Integrate Long-Term Memory Skills with Coding Agent

AgentScope Integration with Long-term Memory Service

AgentScope is an open-source multi-agent platform by Alibaba that provides flexible agent building capabilities. The AnalyticDB for PostgreSQL long-term memory engine can automatically extract key user facts from conversations and perform semantic retrieval through vector similarity.

Integrate AgentScope with Long-Term Memory

QwenPaw Integration with Long-term Memory Service

QwenPaw is a personal AI assistant built on the AgentScope ecosystem. It supports multi-channel access through DingTalk, Lark, WeChat, Discord, and more, with memory evolution and proactive service capabilities. The AnalyticDB for PostgreSQL Long-term Memory Service provides QwenPaw with a cloud-based persistent long-term memory backend that supports cross-device sharing and large-scale semantic retrieval, enabling the AI assistant to "remember" user preferences and historical interactions on any device.

Integrate QwenPaw with Long-Term Memory Service

OpenClaw Integration with Long-term Memory Service

OpenClaw is a multi-channel AI assistant framework that supports local deployment. It can connect to 20+ messaging channels including WhatsApp, Telegram, Lark, and iMessage, and extends memory, skill, and tool capabilities through a plugin mechanism. The AnalyticDB for PostgreSQL Long-term Memory Service can serve as a persistent memory backend for OpenClaw, enabling agents with cross-session and cross-device long-term memory capabilities.

Integrate OpenClaw with the Long-Term Memory Service

Long-term Memory Integration with Coding Agent via MCP

The AnalyticDB for PostgreSQL Long-term Memory MCP Server is a service based on the MCP (Model Context Protocol) that provides AI coding assistants with cross-session persistent memory capabilities. MCP is a standard protocol defined by Anthropic. By registering the adbpgmem MCP Server, AI coding assistants can proactively invoke memory tools during conversations to perform CRUD operations on knowledge.

• Qoder long-term memory integration via MCP

• Integrate Long-Term Memory with Claude Code via MCP

Long-term Memory Integration with Coding Agent via Hooks

AnalyticDB for PostgreSQL Long-term Memory Hooks are a session lifecycle hook integration solution provided by AnalyticDB for PostgreSQL, providing AI coding assistants with seamless cross-session persistent memory capabilities.

• Integrate Long-Term Memory with Qoder via Hooks

• Integrate Long-Term Memory with Claude Code via Hooks

Database Query and Analysis Using AnalyticDB PostgreSQL Query Skill

AnalyticDB PostgreSQL Query Skill is a database query skill designed for AI agents. It enables any AI agent with shell execution capabilities (such as Claude, Qoder, and ClawHub) to connect to AnalyticDB for PostgreSQL via psql, execute read-only queries, and optionally export results as local CSV files.

Use AnalyticDB PostgreSQL Query Skill for database query analysis

Feature

Description

References

PL/pgSQL Functions for Version 7.0

PL/pgSQL (Procedural Language/PostgreSQL) is the default procedural language supported by AnalyticDB for PostgreSQL. The PL/pgSQL implementation in AnalyticDB for PostgreSQL is based on PostgreSQL 12. You can use PL/pgSQL to create user-defined functions (UDFs) for complex data processing and business logic.

PL/pgSQL Functions in version 7.0

Feature

Description

References

Enterprise SSO Access Based on Supabase

AnalyticDB for PostgreSQL is deeply integrated with the open-source Supabase identity authentication system. It supports enterprise access through SAML 2.0 single sign-on (SSO) configuration, meeting the needs of enterprise scenarios, and is especially suitable for building enterprise-grade Coding Agent platforms.

Enterprise SSO integration with Supabase

Data Map for Display Information

In the DataWorks display interface, you can view data such as storage usage, field information, partition information, change records, and DDL information for specified tables in a database. DataWorks refers to this as a data map. AnalyticDB for PostgreSQL provides corresponding query APIs to support the information required by the data map.

Get table metadata for a data map

Recording LLM Calls in LiteLLM via AnalyticDB Supabase

LiteLLM provides a unified interface for calling multiple LLM APIs and writes metadata for each call to external storage through callbacks. AnalyticDB for PostgreSQL Supabase (hereinafter referred to as AnalyticDB Supabase) can serve as the storage target for LLM call logs and billing records. Both successful and failed requests can be recorded for querying, statistics, and troubleshooting in the SQL Editor or Table Editor.

Integrate LiteLLM with AnalyticDB Supabase to log LLM calls

Server-side Connection Pooling (PgBouncer)

If your business scenario involves a large number of short-lived connections or frequent connection creation and destruction, we recommend using the server-side connection pooling feature. This feature significantly reduces the impact on instance performance by reusing database connections, and helps you better control the load and resource consumption caused by establishing and destroying connections on the database server.

Server-side connection pool (PgBouncer)

Supabase CLI Usage

During development, you can use the Alibaba Cloud Supabase CLI to manage the full lifecycle of AnalyticDB for PostgreSQL Supabase (abbreviated as AnalyticDB Supabase) projects from a local terminal, including project creation, database operations, migration management, and edge function deployment. The CLI tool is designed specifically for AnalyticDB Supabase users and has Alibaba Cloud mode enabled by default, so all features are available without additional configuration.

Supabase CLI

AnalyticDB for PostgreSQL AgenticDB Preview

AgenticDB is a data foundation designed specifically for AI agents. It provides one-stop BaaS services and vector retrieval capabilities, with instant start/stop, data sandbox, and context-as-a-service features. As of May 1, 2026, AgenticDB is officially launched and in the preview phase.

• AgenticDB for AnalyticDB for PostgreSQL: private preview

• Auto pause and resume for Supabase

• Data Sandbox Management

• Context service

Feature

Description

References

AI Screenwriting Assistant

Based on built-in GraphRAG (Graph-enhanced Retrieval-Augmented Generation) and Long-term Memory technologies, AnalyticDB for PostgreSQL provides AI-powered creative capabilities for novels, screenplays, and short-form video production. Unlike the traditional "prompt + content generation" approach, the AI Screenwriting Assistant builds an intelligent creation system with character consistency, plot coherence, personalized storytelling, and continuous evolution capabilities.

AI Screenwriter Assistant

Feature

Description

References

Serverless Pro Mode

AnalyticDB for PostgreSQL 7.0 introduces the new Serverless Pro mode. This mode leverages the resource pooling and massive object storage capabilities of cloud-native infrastructure, deeply integrating MPP database technology with Serverless technology to achieve compute-storage separation, second-level elastic scaling, and efficient one-writer-multi-reader capabilities. While fully preserving PostgreSQL core semantics and enterprise-grade stability, Serverless Pro mode is designed to help you significantly reduce total cost of ownership in massive data analysis and high-concurrency read scenarios. ...

• Serverless Pro mode

• Serverless Pro quick start

AnalyticDB for PostgreSQL Serverless Pro mode Reserved Resource Configuration Change

As data volume and computational demands grow dynamically, certain computing resources (such as CPU, memory, and cache space) can become bottlenecks for data processing speed. To support dynamic instance scaling, AnalyticDB for PostgreSQL provides the ability to change instance specifications online.

Upgrade or downgrade reserved resources

AnalyticDB for PostgreSQL Dify Plugin Installation and Authorization

By installing and configuring the Dify plugin, you can directly invoke the document parsing, vectorization, retrieval, and large model inference capabilities of AnalyticDB PostgreSQL within Dify workflows, building end-to-end RAG (Retrieval-Augmented Generation) applications from data ingestion to intelligent Q&A.

Install and authorize the AnalyticDB for PostgreSQL Dify plugin

Feature

Description

References

RDKit

AnalyticDB for PostgreSQL provides the RDKit plugin, which supports chemical molecule computation and chemical molecule retrieval. The plugin supports the mol (molecule type) and fp (molecular fingerprint) data types, and based on these, supports comparison operations, similarity computation (Tanimoto, Dice), and index operations.

RDKit