Vision

Joyint is an ecosystem of terminal-native, Git-native productivity tools for developers. All data lives in .joy/ (or .jyn/) as YAML files, versioned with Git. No database, no lock-in, no server required. AI tools estimate, plan, implement, and review - under capabilities, gates, and budgets.

Why Joyint Exists

Most developer tools treat AI as an afterthought - a plugin, an integration, an API call bolted onto workflows designed for humans alone. Meanwhile, AI is already writing code, reviewing pull requests, and triaging bugs. The gap between what AI can do and what tools allow it to do safely keeps growing.

We believe AI should be a first-class collaborator. That means identity, capabilities, accountability, and limits - the same things you expect from any team member. And we believe the tools that manage this collaboration should live where the code lives: in your terminal, in your Git repo, under your control.

How Joyint Is Different

Most PM tools are built around a web UI and a proprietary database. They own your data, require constant connectivity, and bolt on AI features as an afterthought. Joyint takes a fundamentally different approach.

These are not feature gaps that can be patched over time. They are structural differences in architecture, data ownership, and design philosophy.

Structural Advantages

Git-native vs. proprietary database. When your backlog is YAML in Git, you get branching, diffing, merging, and rollback for free. Feature branches carry their own backlog changes. Code review covers planning changes alongside code changes. There is no synchronization problem because there is no separate system to synchronize with.

Offline-first vs. cloud-dependent. Joyint works without any network connection. No server, no API, no latency. This is not a degraded mode - it is the primary mode. Cloud sync is an optional addition, not a requirement.

Data ownership vs. vendor lock-in. Your data is plain YAML on your filesystem. You can read it, script against it, back it up, or migrate it with standard Unix tools. No proprietary format, no export workflow, no API rate limits.

Open source vs. proprietary. The CLI tools are MIT-licensed. You can audit the code, fork it, extend it, or run it without ever creating an account. The platform is commercial, but the core tools are open.

Supported AI Tools

Joy supports two modes of AI integration. In both modes, AI tools are registered as project members with identity, capabilities, and delegation tokens. Their jobs carry cost budgets.

Tool Mode available

External AI agents use Joy's CLI as a tool. The agent calls joy commands to read, create, and manage items. joy ai init detects installed AI tools and generates the appropriate config files with permissions.

For tools that support agent definitions (GitHub Copilot and Qwen Code), joy ai init generates tool-specific agent files from the member's capabilities - so the AI operates within its granted trust, not from manually written prompts that can drift.

Agent Mode (in the Joyint App and platform)

The app and the platform dispatch work to AI tools and track the results. They are the dispatcher, not the runtime - they prepare context, invoke the tool, and track the outcome including cost. The CLI stays in Tool Mode.

When an AI tool acts on behalf of a human, the event log traces accountability back to the delegating user:

[ai:claude@joy delegated-by:mac@phoenix.org]

Supported Forges

Your Joy data lives in your repository, so every Git forge hosts it. The Joyint platform connects to repositories on these forges and keeps its clone in sync:

Releases use the forge connector's own HTTP client and the same sign in as everything else. Where a release backend is still missing, the connector answers unsupported, so joy release publish pushes the tag and skips the release step with a warning.

Issue Sync Planned

A sync between Joy items and forge issues is planned. It would use a mirror pattern: your local .joy/ directory is the source of truth. Changes flow from local to joyint.com (if connected), and from joyint.com to the external forge. The external forge receives a read-only mirror of relevant data - issues, milestones, and status updates.

This means you can use Joy for planning and governance while your team continues to interact with GitHub Issues, GitLab Issues, or Gitea. ForgeSync bridges the gap without requiring everyone to switch tools at once.

Core Principles

Git-native. All data lives in .joy/ inside your repo. YAML files, versioned with Git, no external service required to start. Your backlog is diffable, branchable, mergeable - just like your code.

Terminal-first, not terminal-only. The CLI is the primary interface. The desktop and web app extend access today, with a TUI and mobile apps planned. But the terminal is where developers work, and that is where Joy meets you.

AI as a first-class collaborator. AI tools estimate, plan, implement, and review. Joy orchestrates the handoff between human intent and AI execution through its AI Governance architecture - five pillars that together form the Trust Model.

Simple by default, powerful when needed. 10 core commands cover 95% of daily use. Complexity lives in flags, not in the command hierarchy.

Your data, your process. No vendor lock-in, no cloud requirement, no forced workflow migration. The process is defined but not enforced: zero gates means zero ceremony, add gates when your team or your auditors need them.

Start solo, scale to team. Works offline for one person. Add a server when you need collaboration. The workflow does not change.

Progressive adoption. Each step makes the previous one more valuable, but none requires the next.

The Trust Model

The Joyint Trust Model is the architecture that governs human-AI collaboration. It is built on five pillars, each answering a fundamental question:

The model scales naturally:

For details on each pillar, see the Solutions page.

Compliance and Standards

AI governance is not just a nice-to-have. Regulatory frameworks are catching up with the reality of AI-assisted development. Joyint is designed to help teams meet these requirements without bolting on compliance as an afterthought.

EU AI Act

The EU AI Act (effective August 2025) requires organizations to document AI decision-making, maintain human oversight, and provide transparency about AI involvement. Joyint's Trust Model addresses this directly:

• Human oversight - Gates and allow_ai flags ensure humans approve critical transitions
• Documentation - The event log records every AI action with delegation chains
• Transparency - AI members have visible identity and capabilities, and their jobs record what they cost

GDPR / DSGVO

Data protection is built in, not bolted on:

• Data sovereignty - Your data stays in your Git repo. No third-party database.
• Encryption - AES-256-GCM with keys on your device. Items and files marked sensitive are unreadable to local AI tools.
• Right to deletion - Plain YAML files in Git. Delete what you need, when you need to.
• Data portability - Your data is always readable YAML. No export tools needed.

SOC 2 / ISO 27001

For organizations pursuing SOC 2 or ISO 27001 certification, Joyint provides:

• Audit trails - Append-only event logs in Git; tamper evidence via Judge is planned
• Access controls - Capability-based permissions for humans and AI
• Change management - Every modification tracked and attributed; signed events are planned
• Independent verification (planned) - Judge will operate as a separate process, reading but never writing

International Compliance

Joyint's architecture - self-hosted deployment, encryption, and local data storage - addresses data residency requirements across jurisdictions.

China - PIPL (Personal Information Protection Law). PIPL requires that personal information collected in China stays within China unless specific conditions are met. Joyint's self-hosted model means data never leaves the infrastructure you control. Qwen Code integration provides a native AI tool option for teams operating under Chinese regulations.

USA - CCPA and SOC 2. The California Consumer Privacy Act and SOC 2 frameworks require clear data handling practices and audit controls. Joyint's event log and capability-based access controls align with these requirements, and independent audit via Judge is planned. FedRAMP considerations are addressed by the self-hosted deployment model, which keeps data within approved boundaries.

Japan - APPI (Act on Protection of Personal Information). APPI mandates appropriate handling of personal information with consent and purpose limitations. Joyint's transparent data model - plain YAML, no hidden processing - supports compliance with APPI's transparency requirements.

Data residency globally. Because Joyint stores all data as local files in your Git repository and encrypts with keys that never leave your device, data residency is determined by where you host your infrastructure, not by where a vendor operates their servers.

Regulated Industries

Finance, healthcare, automotive, and public sector organizations face specific documentation requirements for AI-assisted work. Joyint combines cryptographic identity (Trustship), encryption (Guardianship/Crypt), the event log (Traceability), and job cost tracking (Settlement). Signed events and independent audit through Judge are planned to close the compliance chain.

Progressive Adoption

You never have to buy into the whole ecosystem at once. Each layer builds on the previous one, but none requires the next.

Layer 1: Local CLI. Install Joy or Jyn. Manage your backlog or todos from the terminal. No account, no server, no internet required.

Layer 2: Team sync. Share the repo with your team. Add joyint.com for sync with your forge, the web app, and team chats.

Layer 3: AI governance. Register AI tools as project members. Define what they can do, who delegates to them, and what requires human approval.

Layer 4: Compliance and audit. Encrypt what is confidential with Crypt and run the platform on your own infrastructure. Independent verification through Judge, signed events, and compliance reports are planned.

Why Open Source

We love open source. Not as a marketing strategy, but as a conviction.

We release Joy, Jyn, and Crypt under the MIT license, and Jon and Judge will follow, because we believe good tools should be available to everyone. Solo developers, indie hackers, small teams, open source communities, and enterprise organizations - all with the same chance, without privilege or disadvantage.

Our goal is to strengthen the developer community, protect their data from misuse, and elevate collaboration with AI through professional product management. We want to unleash creative minds and see great new apps built by people everywhere - in the open and inside companies.

Good tools set us free.

How We Fund This

The joyint.com platform is a commercial product. It funds our development and infrastructure.

Your Data, Always

Your data is plain YAML in your Git repo. If you stop using Joyint tomorrow, your data stays right where it is. No export, no migration, no vendor negotiation. That is not a feature - it is a promise.