ContextsBase - Unified AI Knowledge

Product Introduction

1. Definition: ContextsBase is a context infrastructure platform and unified knowledge base specifically designed for AI-powered coding agents. It falls under the technical categories of AI development tools, Model Context Protocol (MCP) servers, and agent memory systems.
2. Core Value Proposition: It exists to solve the persistent memory problem in AI-assisted development by providing a single source of truth for project specifications. Its primary value is enabling coding agents like Claude Code, Cursor, and GitHub Copilot to build consistently and accurately by reading project context—features, business rules, data models, and tests—over the MCP standard at the start of every session.

Main Features

1. Unified Project Memory: A centralized repository where developers document features (F-), business rules (R-), data model entities, test cases (T-*), design tokens, and page edits. This structured knowledge is served to AI agents via an HTTP-based MCP server, ensuring they operate with full project awareness instead of guessing.
2. Iteration-Based Development Workflow: Allows users to group features into iterations (e.g., I-1) and "open" them to connected agents. An agent can then claim, implement, and test features in sequence based on a single instruction, automating the handoff and execution of multi-feature development cycles.
3. MCP (Model Context Protocol) Integration: The core technical mechanism. ContextsBase functions as an MCP server. Developers configure their compatible IDE or agent client (via a .mcp.json file) to connect to the ContextsBase HTTP endpoint with an API token, allowing the agent to pull the latest project context dynamically at the start of each coding session.

Problems Solved

1. Pain Point: The "session-zero" problem where AI coding agents lack persistent memory between developer sessions. This leads to incorrect assumptions about data models, missed business logic, repetitive copy-pasting of specifications, and inconsistent output, forcing developers to constantly re-explain context.
2. Target Audience: Software engineers, engineering managers, and technical leads using AI-assisted development tools (Claude Code, Cursor, Copilot, Windsurf) who work on complex or long-running projects. It is particularly valuable for teams needing consistency across multiple agents or developers.
3. Use Cases: Essential for onboarding new team members (or agents) to a project, maintaining consistency in feature implementation against a defined spec, automating the build-out of a predefined product backlog iteration, and ensuring all automated tests align with documented business rules.

Unique Advantages

1. Differentiation: Unlike simple prompt templates or manually managed documentation, ContextsBase is a structured, machine-readable system integrated directly into the agent's workflow via the open MCP standard. It moves beyond chat history or file-based context, providing a dedicated, updateable "brain" for agents.
2. Key Innovation: Its application of the Model Context Protocol (MCP) to create a persistent, externalized memory layer for coding agents. The innovation lies in treating project knowledge—specs, rules, models—as a first-class, versionable data source that agents can query on-demand, fundamentally changing the agent-human collaboration model from reactive instruction to proactive, informed building.

Frequently Asked Questions (FAQ)

1. What is ContextsBase and how does it work with Claude? ContextsBase is an MCP server that provides a persistent memory for AI coding agents. You write your project's specs and rules in ContextsBase, then configure Claude Code via its MCP settings to connect to it. At the start of a session, Claude pulls this context, allowing it to build features with full knowledge of your data model and business rules.
2. Is ContextsBase free to use? Yes, ContextsBase offers a free forever plan that includes one project, support for up to 100 features, 3 team members, and integration with any MCP-compatible agent, which is sufficient for individual developers or small projects.
3. What is MCP and do I need to be an expert to use ContextsBase? MCP (Model Context Protocol) is an open standard developed by Anthropic for tools to provide context to AI models. You do not need MCP expertise; using ContextsBase involves copying a provided configuration block into your agent's settings (like .mcp.json) and adding your API token.
4. Can ContextsBase work with GitHub Copilot? Yes, ContextsBase works with any AI coding assistant that supports the Model Context Protocol (MCP). This includes GitHub Copilot when used within an MCP-compatible editor or client that can be configured to connect to external MCP servers like ContextsBase.
5. How does ContextsBase improve AI agent accuracy? By serving a structured, authoritative source of truth over MCP, it eliminates an agent's need to guess. The agent has direct access to the correct data model field names (e.g., customer.plan_id), asserted business rules for tests, and exact feature specifications, drastically reducing implementation errors and inconsistencies.