OpenAgile.AI

Fast to Generate. Impossible to Maintain. AI closes the gap between idea and code. It doesn't close the gap between code and understanding. Without traceable intent, versioned context, and enforced structure, you're not building a system — you're accumulating it. OpenAgile.AI is the engineering-discipline layer that sits above any coding agent and enforces specifications before generation, validation before execution, and traceability throughout.

Two ways to use it

• Free for Solos
  Solo developers, side projects, learning, academic and research use. Same software, no accounts, runs entirely on your machine.
• Licensed for Business
  Required for companies, internal tools, client and agency work, commercial products, and production deployments. Same software, commercial license from BWS.

Both tracks are governed by the OpenAgile.AI End-User License Agreement.

How it works

OpenAgile.AI is guided by the Agile Vibe Coding Manifesto (CC BY-SA 4.0). Four named values frame how the tool turns intent into shipped code:

• Accountability over anonymous generation
• Traceable intent over opaque implementation
• Discoverable domain structure over scattered code
• Human-readable documentation over implicit knowledge

The full 12 principles are public — accountability, traceable intent, deterministic verifiability, and explicit context are treated as engineering requirements, not aspirations. They are enacted through a structured ceremony pipeline:

• Sponsor Call
  Capture intent, scope, constraints, and success criteria for the project.
• Sprint Planning
  Decompose the project into epics, stories, and atomic tasks — each with its own scoped file and interface context.
• Seed
  Auto-fix specifications via 342 micro-checks across 15 engineering perspectives — semantic deduplication, cross-reference validation, deterministic scoring.
• Run
  Execute tasks in parallel via isolated git worktrees, with multi-LLM auto-fallback across Claude, Gemini, OpenAI, and local models.
• Verify & Merge
  Behavioural and end-to-end tests gate every merge; failures trigger fault diagnosis and automatic fix cycles.

Features

• Hierarchical Decomposition
  Project → epic → story → task, each carrying its own scoped context window so agents only see what's relevant.
• Structured Specs, Not Prompts
  Every task spec defines exactly which files and interfaces are in scope — no ambient prompts, no implicit context.
• Validate Before You Generate
  342 domain checks catch errors at the planning layer, where they're cheapest to fix. A bad plan found in 30 seconds beats a bad feature found in production.
• Bidirectional Traceability
  Code ↔ acceptance criteria ↔ story ↔ epic. Root-cause any change in seconds; prove provenance to auditors on demand.
• Autopilot
  Chain ceremonies end-to-end with parallel execution and watchdog timers — for teams that want hands-off shipping with guardrails.
• Verify & Auto-Fix
  Behavioural and E2E tests validate acceptance criteria; failures trigger automatic fault diagnosis and fix cycles.
• Safe Merge
  Worktree branches with conflict resolution, post-merge test gates, and automatic rollback on regression.
• Multi-LLM Support
  Claude, Gemini, OpenAI, Xiaomi MiMo, LM Studio and Ollama, with automatic provider fallback.
• Local Kanban Board
  Real-time WebSocket project visualisation at localhost:4174. No cloud, no accounts, no telemetry.
• Compliance-Ready by Design
  Audit trails, ownership clarity, and risk visibility built for SOC 2, ISO 27001, and the EU AI Act.

Built for regulated work

Your code must be auditable. AI doesn't change that. OpenAgile.AI was designed from day one for fintech, healthtech, and govtech teams that need to prove what was built, why, and by whom — without sacrificing AI-driven velocity. Bidirectional traceability and 342-check validation aren't bolt-ons; they're built into the spec layer.

Status

Alpha — Available now. Evolving fast. The spec and planning layer is ready and in daily use. Code generation ceremonies are still preliminary. Available today: Sponsor Call, Sprint Planning, Seed, Autopilot, local Kanban board, multi-LLM support. Coming next: Run, Commit, and Verify ceremonies, an integrated code editor, initialisation on existing projects, and team collaboration.