repoprompt/repoprompt-ce

What it solves

RepoPrompt CE addresses the challenge of "context engineering" for AI coding agents. It prevents agents from acting blindly by assembling the exact, focused codebase information (files, structure, and diffs) they need to perform a task, while allowing the user to review and refine that context before it is sent to the model.

How it works

It operates as a native macOS application and agent orchestrator. It uses a bundled Model Context Protocol (MCP) server to allow MCP-compatible clients and CLI agents to search repositories, inspect files, and curate context. The app can manage multi-root workspaces and worktrees, enabling agents to explore a repository and identify relevant files within a specific token budget.

Who it’s for

Developers using AI coding agents and CLI-based AI tools who want more control over the context provided to their models to improve accuracy and reduce token waste.

Highlights

• Context Engineering: Combines file trees, line slices, CodeMaps, and Git diffs into dense prompts.
• Agent Orchestration: Coordinates CLI-backed coding agents directly from a macOS interface.
• MCP Integration: Includes a bundled MCP server to connect external tools to the repository context.
• Reviewable Handoffs: Allows users to inspect and edit the context before it is handed off to an AI tool.
• Multi-root Support: Works across multiple related repositories and documentation folders simultaneously.