# ContextsBase

> Replace loose markdown notes with an agent-native project workspace. Structure specs, schemas, and test rules once—Claude Code, Cursor, and Copilot pull them over MCP and build without guessing.

ContextsBase is a single-source context platform (web app) that manages unified project knowledge for AI agents. Served seamlessly over MCP, it holds your specs, rules, data models, workflows, and guidelines in one place—so your agents always build with exact context.

While built natively to power coding agents (Claude Code, Cursor, Copilot, Windsurf), ContextsBase isn't limited to web development. Whether you're building backend APIs, mobile apps, systems software, or managing marketing campaigns and brand engines, ContextsBase acts as the central source of truth that your agents read from and report back to.

## What it holds

- Features (F-n): specifications with functionality, business rules, and edge cases. Draft until published; agents only read published features.
- Entities: the domain data model with fields and foreign keys.
- Flows: decision trees for user journeys and branching logic.
- Test cases (T-n): plain-English Given/When/Then cases that assert business rules. Agents turn them into Playwright tests and record the result.
- Iterations (I-n): ordered lists of features that agents pull from one step at a time, with atomic claims.
- Theme: colors, typography, radius, and spacing tokens designed visually with a live preview and published for agents to read.
- Page edits: click-to-edit changes to a live page, saved as edits the agent applies to the real code.

## How agents connect

- Native MCP server over JSON-RPC 2.0 with token scopes for read, write, and records.
- Tools: list_iterations, next_iteration_item, get_feature_spec, record_test_automation, record_feature_implementation.
- Every implementation record carries a fingerprint of the spec it was built from. Editing the spec later marks the record Outdated.

## Pricing

- Free: 1 project, 100 features, 3 members.
- Premium (coming soon): $5 per month. Unlimited projects, features, and members.

## Guides

Index of every guide: https://contextsbase.com/guides

### Agent setup

How to give a coding agent your project context, and where each method breaks.

- MCP Server for Cursor: Project Context & Agent Backlog: Connect ContextsBase to Cursor over MCP. Serve specs, schemas, and tests directly into Cursor Composer and Agent mode without bloating .cursorrules. https://contextsbase.com/mcp/cursor
- MCP Server for Claude Code: Specs, Tests & Autonomous Iterations: Connect ContextsBase to Claude Code CLI over MCP. Turn terminal sessions into autonomous backlog execution with persistent specs and test reporting. https://contextsbase.com/mcp/claude-code
- How to Give Claude Code Context About Your Project: Give Claude Code project context with CLAUDE.md, memory files, or an MCP server. Where each one breaks, and how to serve specs, data model, and tests. https://contextsbase.com/guides/claude-code-project-context
- How to Give Cursor Context About Your Project: Four ways to give Cursor project context, from .cursorrules to an MCP server with specs, data model, and tests. What works, what breaks, when to upgrade. https://contextsbase.com/guides/cursor-project-context
- MCP server for coding agents: Specs, business rules, data model, and tests served to Claude Code, Cursor, Copilot, or Windsurf. Tools, scopes, and how records stay in sync with the spec. https://contextsbase.com/mcp-server-for-coding-agents
- The Complete Guide to Model Context Protocol (MCP): Architecture & Real-World Use Cases: An in-depth guide to Model Context Protocol (MCP) architecture, JSON-RPC primitives, transports (stdio & HTTP SSE), tools, resources, prompts, and real-world developer use cases. https://contextsbase.com/guides/mcp-architecture-and-use-cases
- 10 Best MCP Servers Every Developer Should Use in 2026: A curated list of the top Model Context Protocol (MCP) servers for codebase context, database inspection, API testing, memory, and live execution in Claude Code, Cursor, and Windsurf. https://contextsbase.com/guides/best-mcp-servers-2026
- How to Connect MCP Servers to Claude Code & Cursor Step-by-Step: Step-by-step tutorial on connecting stdio and HTTP SSE MCP servers to Claude Code and Cursor. Command lines, config JSON examples, header auth, and troubleshooting. https://contextsbase.com/guides/connect-mcp-claude-code-cursor

### Comparisons

How ContextsBase differs from the tools teams already use, and when you want both.

- Mintlify Alternative for Coding Agents: Specs, Not Docs: Mintlify publishes docs for your users' agents. ContextsBase is the spec, data model, and tests your own coding agents build from. Which one you need, and when both. https://contextsbase.com/compare/mintlify-alternative
- Notion Alternative for Developers and AI Agents: Notion holds specs for people. ContextsBase holds the specs, rules, data model, and tests your coding agent reads over MCP. What changes when the reader is an agent. https://contextsbase.com/notion-alternative-for-developers
- Why MCP is Replacing Custom Plugins for AI Coding Assistants: Why closed plugin ecosystems are giving way to open Model Context Protocol (MCP) servers. Comparing plugin architecture vs open standards, security, multi-client interoperability, and state management. https://contextsbase.com/guides/why-mcp-is-replacing-plugins
- Jira for AI Agents: Write Features, Not Tickets: Replace tickets with features that Claude Code, Cursor, and Copilot read over MCP, build in order, and record. No assignees, no sprints. https://contextsbase.com/jira-for-ai-agents

## Links

- Landing page: https://contextsbase.com/
- Guides: https://contextsbase.com/guides
- Support: https://contextsbase.com/support
- Sign in: https://app.contextsbase.com/login
- Create a free project: https://app.contextsbase.com/register