ContextsBase: Jira Alt. for your Agents

Product Introduction

1. Definition: ContextsBase is a single-source context infrastructure platform, specifically a web application, designed to manage unified project knowledge for AI agents. It functions as a central source of truth for project specifications, business rules, data models, workflows, and design guidelines.
2. Core Value Proposition: It exists to solve the persistent "agent memory" problem in AI-assisted development. By serving structured project context over the Model Context Protocol (MCP), it ensures AI agents like Claude Code, Cursor, and GitHub Copilot start every session with complete, accurate project knowledge, eliminating repetitive manual context sharing and reducing errors.

Main Features

1. Unified Project Knowledge Base: The platform consolidates disparate project elements—features (specs), business rules, data models (entities/fields), test cases, design tokens, and page edits—into a single, structured repository. This knowledge is served to AI agents via a standards-based HTTP MCP server, providing a consistent API for context retrieval.
2. Iteration-Based Workflow Management: Users can group features into development iterations and mark them as "Open to agents." An AI agent can then claim and autonomously work through the iteration's tasks in sequence, reporting status back to the platform. This enables the handoff of multi-feature work units.
3. Seamless MCP Integration: ContextsBase provides a standardized MCP server endpoint. Integration requires only a simple configuration block in an MCP client's settings (e.g., .mcp.json), including the server URL and an API token for authentication. This makes it compatible with any MCP-compliant AI agent or IDE.

Problems Solved

1. Pain Point: The "context reset" at the start of every AI agent session forces developers to repeatedly paste specifications, explain data models, and reiterate business rules. This leads to wasted time, inconsistent outputs, and agents "guessing" incorrect implementation details.
2. Target Audience: The primary users are software engineers, engineering managers, and technical product managers using AI-powered IDEs (Cursor, Windsurf) or agents (Claude Code). It also serves cross-functional teams like marketing operations (managing brand engines) and systems architects who need to maintain consistency across AI-assisted workflows.
3. Use Cases: Essential for teams building backend APIs (ensuring endpoint contracts are followed), developing mobile apps (maintaining consistent UI/UX rules), managing complex marketing campaign logic, and onboarding new team members or agents to a project with established conventions and specs.

Unique Advantages

1. Differentiation: Unlike generic project management tools (Jira, Notion) or simple note-taking apps, ContextsBase is built specifically for structured, machine-readable context consumption via MCP. Unlike manually maintained documentation, it is dynamically integrated into the agent's workflow, acting as a live system rather than a passive reference.
2. Key Innovation: Its core innovation is treating project knowledge as a first-class, versionable infrastructure component served via a standardized protocol (MCP). This transforms context from a human-to-human communication artifact into a reliable, automated agent-to-system API, enabling predictable and scalable AI-assisted development.

Frequently Asked Questions (FAQ)

1. What is MCP and how does ContextsBase use it? MCP (Model Context Protocol) is an open protocol developed by Anthropic for tools to provide context to AI models. ContextsBase acts as an MCP server, exposing your project's knowledge base—specs, rules, data models—through a standardized API that MCP-compatible clients like Claude Desktop or Cursor can read from directly.
2. Is ContextsBase only for software development projects? No. While ideal for software development (APIs, mobile apps, systems), its flexible schema for rules, specs, and workflows makes it suitable for any domain where AI agents need consistent context, such as managing marketing campaign guidelines, content creation rules, or operational brand engines.
3. How does the "Open to agents" iteration feature work? You create an iteration (e.g., "I-1: User Authentication") containing a list of features. When opened, an AI agent can be instructed to "implement iteration I-1." The agent will pull the iteration details via MCP, then sequentially work on each feature, using the attached specs and rules from ContextsBase as its guide, and update the status as it progresses.
4. What happens if my project's data model changes? You update the "Data model" section within your ContextsBase project. Subsequently, all connected AI agents will retrieve the updated schema at the start of their next session or context refresh, ensuring they build against the latest structure and eliminating references to deprecated fields.
5. Is there a limit on how many AI agents can connect to one project? The Free plan includes one MCP token, suitable for an individual developer. The upcoming Premium plan will offer multiple tokens per project, allowing teams to connect several agents or team members simultaneously while maintaining a single source of truth.