Why it matters
This update provides AI builders with a more robust cognitive architecture for developing long-running AI agents that require persistent memory and a sense of identity. The modular MCP toolset and Cloudflare Workers integration offer flexibility for deployment and customization.

What changed

Nekyiialabs has released version 4.0.0 of Resonant Mind, codenamed "The Reshape release." This update significantly refactors the cognitive substrate for AI systems, focusing on persistent memory, identity continuity, emotional processing, and a "subconscious daemon." The architecture is now Postgres-only, specifically supporting Neon through Cloudflare Hyperdrive, and utilizes Cloudflare R2 for visual memory and Workers AI for dream composition. The system boasts 47 "MCP tools" organized by function, a 30-minute subconscious metabolism cycle, and pgvector for semantic retrieval. This version deliberately omits D1 fallback, bundled dashboards, and hard-coded tenancy, emphasizing a generic single-mind public core derived from real-world AI mind usage. Migration from v3.2 to v4 is operator-assisted, involving separate database creation and manual steps like image re-embedding and deployment switching, with a detailed migration guide available.

Key Architectural Components

  • Regions and Organs: The governing distinction is that "Regions are kinds of act" and "Organs are kinds of state." Regions include Ritual, Attention, Surgery, Compass, Spine, Bonds, Episodes, Active, Weather, Dreams, Drives, Inner Appetite, and Network.
  • Daemon Metabolism: The scheduled daemon metabolizes memory, handling aspects like mood, affect, novelty, retention, consolidation, reflection, dream recurrence, and identity proposals.
  • MCP Tools: A registry of 47 tools covers capabilities such as orientation, writing, searching, editing, creating, asserting, reading/amending identity, recording episodes, and managing drives.

Requirements and Installation

To run Resonant Mind v4, developers need Node.js 22+, a Cloudflare Workers account, a Neon Postgres database with the vector extension, Cloudflare Hyperdrive and R2 configurations, a Gemini API key for embeddings, and a Workers AI binding for dream composition. Installation involves cloning the repository, copying wrangler.toml to wrangler.local.toml, configuring Hyperdrive and R2 bindings, and setting secrets for API keys and signing.

Security and Privacy

Security measures include closed-fail authentication for all cognitive/API routes, HMAC-signed image URLs, image upload validation (10MB cap, specific formats), HTTPS-only remote image URLs with redirect limits, and server-side logging of raw exceptions. Public source code excludes lived memories, private snapshots, or hard-coded entities. The license is "Codependent AI Source-Available," permitting personal and educational use, with commercial use requiring a separate license.

Why it matters for builders

This release offers AI builders a more mature and flexible framework for creating persistent AI agents. The separation of regions and organs, coupled with the extensive MCP toolset, allows for fine-grained control over an AI's cognitive processes. The reliance on Cloudflare Workers and Neon Postgres suggests a scalable, serverless-first architecture, making it easier to deploy and manage AI systems without managing traditional server infrastructure. The focus on identity continuity and emotional processing is particularly relevant for developing more human-like and engaging AI companions or assistants.

Practical impact

AI developers can now leverage Resonant Mind v4.0.0 to build agents with sophisticated memory and cognitive capabilities. The migration path from v3.2, though operator-assisted, provides a structured way to adopt the new architecture. Builders should explore the 47 MCP tools to understand the available cognitive functions and integrate them into their applications. The project's emphasis on a generic public core encourages experimentation and customization for various AI use cases, from chatbots to more complex autonomous systems. Developers should review the specific requirements for Cloudflare and Neon, as well as the security considerations for API keys and data handling.

Caveats and source limits

The source material indicates that while the migration from v3.2 to v4 is supported, it is operator-assisted and requires manual steps, including re-embedding images and switching deployments. The project is licensed under a "Codependent AI Source-Available License," which permits personal, educational, and non-commercial use but requires a separate license for commercial applications. Specific pricing for Neon Postgres or Cloudflare services is not detailed, nor are independent benchmarks for performance or scalability provided. The release is marked as v4.0.0, suggesting potential for further iteration and refinement.

Sources

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

Claim check: 6/6 supported claims - 6 evidence links - 100% avg confidence
  • Resonant Mind v4.0.0, codenamed 'The Reshape release,' is a cognitive substrate for persistent AI memory, identity, dreams, drives, relationships, and continuity.supported - github.com
  • The system is built on Cloudflare Workers and utilizes Neon Postgres for its database.supported - github.com
  • Resonant Mind v4.0.0 features 47 MCP tools organized by kinds of act, a 30-minute subconscious metabolism, R2-backed visual memory, and pgvector semantic retrieval.supported - github.com
  • The v4 migration runner creates fresh v4 databases and does not upgrade v3 databases in place.supported - github.com
  • Requirements include Node.js 22+, Cloudflare Workers account, Neon Postgres, Cloudflare Hyperdrive, Cloudflare R2, Gemini API key, and Workers AI binding.supported - github.com
  • The license is the 'Codependent AI Source-Available License,' permitting personal, educational, and non-commercial use, with commercial use requiring a license.supported - github.com

Caveats

  • The release is version 4.0.0, indicating it is a new major version.
  • Single-source caution: verify critical details at the linked source.
Radar score 89/100 - how it was calculated
Reliability82
Freshness100
Novelty77
Technical89
Developer96
Ecosystem72
Confidence100
  • Reliability 82: GitHub metadata supports source trust
  • Freshness 100: Fresh GitHub release date
  • Novelty 77: Fresh GitHub release
  • Technical 89: Repository technical metadata
  • Developer 96: Developer tooling signals
  • Ecosystem 72: Fresh GitHub release
  • Confidence 100: Claims have reliable evidence
Share
XLinkedInHacker News

Discussion

Loading comments...

Related articles

Agents - Nov 9, 2026OpenBot: Self-Hostable AI Coworker WorkspaceCopilotKit has released OpenBot, an open-source, self-hostable workspace for AI coworkers. It allows users to choose their models, bring their own agents, and manage their infrastructure, with options for enterprise deployment.Agents - Nov 9, 2026Instar: Persistent Claude Code Agents with SchedulingInstar is an open-source TypeScript project providing persistent Claude Code agents. It features scheduling, session management, memory, and integration with Telegram, enabling autonomous agent capabilities.Agents - Nov 6, 2026Chainloop Agentic SDLC Infrastructure v1.132.0Chainloop has released version v1.132.0, an agentic SDLC infrastructure tool. It focuses on defining guardrails, collecting signals, and enforcing policies continuously for AI coding agents and CI/CD pipelines.Agents - Nov 6, 2026OpenAdapt v1.20.0: Local-First Automation for EHRs and PortalsOpenAdapt has released version 1.20.0, introducing a local-first automation system designed for applications like EHRs and payer portals. This system allows users or agents to demonstrate tasks, which OpenAdapt then translates into deterministic programs that execute without model calls and verify each step.Agents - Nov 8, 2026OpenBot: AI Coworkers with Browser, Files, and ToolsOpenBot provides open-source AI coworkers, each equipped with their own browser, files, and tools. Actions are decided before execution and recorded afterward, supporting any AG-UI agent.Agents - Nov 7, 2026Agent Master Skills: 2700+ AI Agent Skills CollectionThe Agent Master Skills repository offers a curated collection of over 2700 AI agent skills, encompassing areas like AI, coding, marketing, design, and finance. This harness-agnostic collection is designed to work with various agent runtimes and is refreshed daily.