Loading...
The Problem
AI-generated code works, but context disappears. Three months later, your own code looks foreign.
The Solution
Context Mesh makes context the primary artifact. Code becomes its manifestation, not the other way around.
A process framework for AI-First development that treats context as the primary creation, with code as its manifestation, enabling sustainable AI-assisted development.
Built for effective AI-human collaboration with clear benefits that transform your development workflow.
Context Mesh is designed for simplicity and easy adoption. Three essential steps that preserve context throughout the development lifecycle.
Defines what you want to build and why, creates feature/bug intents, makes technical decisions, creates initial living context, and identifies patterns.
Lead intent capture, validate intent clarity, approve initial context, align stakeholders.
Assist in structuring intent, suggest context organization, generate context from prompts, explain before executing.
✅ Context as Primary Creation - Context is created first | ✅ Intent-Driven Architecture - Intent guides everything
Context Mesh uses a simple, clear directory structure that's easy to navigate and maintain. Works with any tools or no tools at all.
Organized structure that both humans and AI can easily understand and navigate.
Add subdirectories as needed. The structure grows with your project.
Context is versioned (Git), searchable, and always up-to-date. Full traceability from intent to code.
12345context/├── intent/│ └── feature-user-auth.md # What + Why + Acceptance criteria└── decisions/ └── 002-auth.md # How (technical approach) + RationaleStart simple. Add complexity only when needed.
Understanding when to create vs update, naming conventions, and file management.
Rule of Thumb: Ask yourself 'Is this a new scope or evolution of existing scope?'
| Situation | Action | Why |
|---|---|---|
| New feature | Create feature-*.md | New scope |
| Update existing feature | Update feature-*.md | Same scope (Git preserves history) |
| New decision | Create decisions/003-*.md | New scope |
| Add outcomes to decision | Update decisions/002-*.md | Same scope |
When a feature is removed or a bug is resolved, mark as deprecated/resolved but do NOT delete. Keep files for history and traceability.
Include at the end of every context file:
1234## Status- **Created**: YYYY-MM-DD (Phase: Intent/Build/Learn)- **Status**: [Status Type]- **Updated**: YYYY-MM-DD (Phase: Intent/Build/Learn) - [reason] (optional)Status Types:
Context Mesh adapts to your needs. Different approaches are equally valid:
When to Use:
Exploratory projects, unclear requirements, quick prototypes
How:
Start with basic intent, use AI to expand context, learn as you go
When to Use:
Large projects, clear requirements, team projects
How:
Include decisions, patterns, DevOps planning in Step 1, create complete context before Build
When to Use:
Legacy codebases, taking over projects, refactoring
How:
Use AI to analyze existing code, generate context from codebase, then follow normal workflow
Context Mesh defines clear roles for human and AI at each step, ensuring effective collaboration and maintaining human control.
Human leads, AI assists. Define goals and validate context together.
AI builds, human supervises. Generate code with full context awareness.
AI analyzes, human validates. Capture learnings and update context.
AI explains what it will do
You review and approve
AI executes with approval
Context Mesh implements five fundamental principles that guide AI-First development.
Context Mesh integrates seamlessly with Scrum, Kanban, DevOps, or your own development process.
Context Mesh uses Definition of Done at the technical/feature level only - criteria that every feature implementation must meet before being considered complete.
Acceptance Criteria:
Acceptance Criteria (in feature-*.md files) - What the feature needs to do functionally (e.g., 'User can login', 'Data is saved')
Definition of Done:
Definition of Done - Process criteria that must be met during implementation (e.g., 'Tests passing', 'Code reviewed', 'Context updated')
Context Mesh adapts the AGENTS.md standard (see agents.md) to work seamlessly with our context structure. AGENTS.md acts as a router that directs AI agents to Context Mesh files.
An open format used by over 20,000 open-source projects (see agents.md). Context Mesh adapts this standard to work with our context structure.
Learn about AGENTS.md standardOur AGENTS.md focuses on routing AI agents to Context Mesh files (intent, decisions, patterns) rather than duplicating context. Keep it succinct - it should primarily indicate where to find context.
Important: AGENTS.md should be kept updated to reflect changes in the living context. New features, decisions, and patterns should be added to the appropriate sections.
Security is built into Context Mesh from the ground up. Consider security at every step.
Get started in minutes with our quick start guide. Just copy, paste, and let AI create your context structure.