Why it matters
This release offers builders enhanced control and safety for database changes through AI-driven prechecks and robust governance workflows. The platform's ability to integrate with AI coding agents and its cross-engine rollback capabilities are particularly valuable for teams seeking to automate and secure their database CI/CD pipelines.

What changed

Yearning v4 represents a complete architectural overhaul of the open-source Yearning SQL audit and database governance platform. The new version is built as a single-binary Go modular monolith, paired with a React 19 + shadcn/ui web workspace and a PostgreSQL metadata plane. It eliminates legacy RPC services and introduces several key features:

Core Features:

  • Zero-Leak AI SQL Precheck & Submission Gate: This engine performs a deterministic and AI-driven two-stage precheck before any DDL/DML change is submitted. It lexes SQL batches (up to 100,000 statements or 32 MiB), computes canonical and dialect fingerprints, and collapses homologous statements. The AI Review Agent inspects a single, anonymized SQL representative per group. It includes 5 read-only agent tools (check_syntax, describe_table, list_indexes, table_stats, safe_explain) that inspect schema metadata and execution plans without mutating data. The platform supports dual-track AI privacy isolation: a default track for cloud OpenAI-compatible APIs with in-memory anonymization, and a san track for local self-hosted models via vLLM/Ollama with strict isolation. It also manages custom prompt skills, enterprise knowledge bases, and enforces a critical submission gate for policy violations or invalid statements. A global daily AI cost budget with failover across providers is also included.
  • Frozen Multi-Stage Approvals & Execution Control: Users can configure change_review flows with 1..N serial datasource stages (e.g., Staging to Production) and 1..10 serial approval steps per stage, along with explicit execution actors. Rules, stages, datasources, approvers, executors, SQL SHA-256 hash, and AI review snapshots are frozen server-side upon submission, preventing any role, including admin, from editing, delegating, skipping, or overriding submitted orders. SQL is encrypted at rest (AES-256-GCM) and omitted from default responses; viewing plaintext triggers an audited reveal with a traceable watermark. Execution is safe, with approval not auto-executing; frozen executors trigger immediate execution or create a scheduled run. A single-datasource execution lease prevents concurrent conflicting changes. The platform includes built-in gh-ost for native MySQL 5.7/8.0 ALTER TABLE changes, running in-process with tunable safety parameters and real-time progress. Fail-closed result_unknown recovery prevents blind replays of SQL during network drops.
  • Cross-Engine Automatic SQL Rollback: Yearning v4 introduces a zero-Binlog/zero-WAL cross-engine Abstract Syntax Tree (AST) and before-image rollback architecture for both MySQL-compatible and PostgreSQL-compatible engines. It captures exact row before-images during execution (up to 100,000 rows per statement/order), automatically aborting and rolling back the forward transaction if this limit is exceeded. Structural DDL and gh-ost snapshots are captured for pre/post schema metadata.

The platform supports MySQL, PostgreSQL, TiDB, OceanBase, and PolarDB. It is built with Go and React 19, utilizing PostgreSQL for its metadata plane.

Why it matters for builders

Yearning v4's release provides developers and DBAs with a significantly more secure and automated approach to managing database changes. The integration of AI for pre-checks, especially with options for both cloud and local models, allows for proactive identification of risks and policy violations before they reach production. The immutable governance workflows and cross-engine rollback capabilities reduce the fear of accidental data loss or system downtime, enabling faster iteration cycles.

Practical impact

Builders can leverage Yearning v4 to integrate AI coding agents more safely into their database workflows, ensuring that AI-generated SQL adheres to organizational policies and security standards. The platform's support for multiple database engines and its robust governance features make it suitable for enterprise environments seeking to standardize their database CI/CD processes. Developers can explore the live demo at demo.yearning.io or consult the documentation at v4.yearning.io for installation and configuration guidance. The project is available on GitHub, encouraging community contributions and feedback.

Caveats and source limits

The provided source text details the features and architecture of Yearning v4 but does not include specific benchmark results for the AI precheck performance or the rollback capabilities. Information regarding pricing for enterprise licensing or commercial support is also not detailed, though an email address for inquiries (henry@yearning.io) is provided. The release status is indicated as a "fresh release" with the latest version being v4.0.0-rc.5, suggesting it may still be in a release candidate phase.

Sources

Written with AI assistance from the linked sources; every claim below was checked against them automatically. How we produce articles.

Claim check: 12/12 supported claims - 12 evidence links - 100% avg confidence
  • Yearning v4 is a ground-up architectural rebuild of the open-source Yearning SQL audit and database governance platform.supported - github.com
  • Yearning v4 is built as a single-binary Go modular monolith paired with a React 19 + shadcn/ui web workspace and a PostgreSQL metadata plane.supported - github.com
  • Yearning v4 introduces a zero-leak AI SQL Review Engine, cryptographically frozen multi-stage governance, built-in in-process gh-ost Online DDL, binlog-free cross-engine rollback, governed read-only queries, and Top-SQL performance diagnostics.supported - github.com
  • Yearning v4 supports MySQL, PostgreSQL, TiDB, OceanBase, and PolarDB.supported - github.com
  • The AI Precheck engine collapses up to 100,000 statements or 32 MiB per SQL batch into fingerprint groups, inspecting only one anonymized SQL representative per group.supported - github.com
  • The AI Precheck includes 5 read-only agent tools: `check_syntax`, `describe_table`, `list_indexes`, `table_stats`, and `safe_explain`.supported - github.com
  • Yearning v4 offers dual-track AI privacy isolation: `default` track for cloud OpenAI-compatible APIs with in-memory anonymization, and `san` track for local self-hosted models via vLLM/Ollama.supported - github.com
  • Yearning v4 supports `1..N` serial datasource stages and `1..10` serial approval steps per stage for change review flows.supported - github.com
  • SQL changes are encrypted at rest with AES-256-GCM and omitted from default responses, with viewing plaintext triggering an audited reveal carrying a traceable watermark.supported - github.com
  • Yearning v4 includes built-in `gh-ost` for in-process MySQL 5.7/8.0 `ALTER TABLE` changes.supported - github.com
  • Yearning v4 features a zero-Binlog/zero-WAL cross-engine AST and before-image rollback architecture for both MySQL-compatible and PostgreSQL-compatible engines.supported - github.com
  • The rollback feature captures exact row before-images during execution, up to 100,000 rows per statement and per order, with automatic rollback if the limit is exceeded.supported - github.com

Caveats

  • Single-source caution: verify critical details at the linked source.
Radar score 87/100 - how it was calculated
Reliability82
Freshness100
Novelty77
Technical82
Developer96
Ecosystem72
Confidence96
  • Reliability 82: GitHub metadata supports source trust
  • Freshness 100: Fresh GitHub release date
  • Novelty 77: Fresh GitHub release
  • Technical 82: Repository technical metadata
  • Developer 96: Developer tooling signals
  • Ecosystem 72: Fresh GitHub release
  • Confidence 96: Claims have reliable evidence
Share
XLinkedInHacker News

Discussion

Loading comments...

Related articles

Agents - Nov 17, 2026ESP32 iPhone Agent enables AI control of iPhonesThe ESP32 iPhone Agent project allows AI models to control an iPhone using a low-cost ESP32 microcontroller. It enables AI to see the screen via AirPlay, interact using USB HID for touch input, and process audio for voice commands, all without requiring a Mac, developer account, or jailbreaking the device.Agents - Nov 17, 2026BuilderIO Agent Native: Framework for Agentic AppsBuilderIO Agent Native is an open-source TypeScript framework designed for building agentic applications. It recently released version v0.1.298, indicating active development and a focus on providing tools for AI agents.Agents - Nov 16, 20260security: Open-Source AI Cybersecurity Harness0security is an open-source framework for building continuous security agents. These agents can identify, prove, fix, and monitor vulnerabilities across code, dependencies, infrastructure, and AI agents, running on user-provided infrastructure with chosen LLMs.Agents - Nov 20, 2026PraisonAI v4.7.26: Offline AI Agents & Enterprise HandoverPraisonAI has released version 4.7.26 of its open-source Python agent framework, enabling fully offline AI agent execution. The update also introduces an early access feature for highly regulated teams, allowing AI tasks to be handed over to self-hosted models with a verifiable audit trail.Agents - Nov 18, 2026Rome OS: Open-Source Agent OS for Recursive AgentsRome OS is an open-source operating system for recursive AI agents, designed to scale the agent's environment rather than just model size. It offers a collaborative, guardrailed space for human-agent interaction, enabling agents to build their own workflows and improve over time.Agents - Nov 18, 2026Weaxs Stock Analysis Plugin for AI AgentsWeaxs has released a stock research toolkit plugin designed for AI agents, offering 73 tools and 21 skills for market analysis, screening, and backtesting across multiple global markets. This plugin is compatible with various AI agent frameworks, including Pi Agent, Hermes, OpenClaw, and DeepSeek Harness.