mini-kode is an open-source terminal-based AI coding assistant developed by minmaxflow. Built as a lightweight, modular CLI coding agent, it provides automated code exploration, file reading, and task execution directly from the command line while incorporating interactive UI components for managing agent actions. In standard day-to-day operation, mini-kode interacts with the user via a terminal interface (TUI/CLI). The agent interprets natural language tasks, scans and inspects repository files using dedicated filesystem tools (like fileRead), executes terminal commands, and integrates external capabilities via Model Context Protocol (MCP) servers. An interactive permission subsystem prompts users to approve or decline file access paths and shell commands before execution. What distinguishes mini-kode is its explicit focus on permission granularity and transparent MCP tool result visualization. Through dedicated TUI components like the PermissionSelector and MCPResultView, it enforces strict discriminated-union permission options across paths and commands, shutting down external MCP processes gracefully upon exit. mini-kode is released as open-source software under a permissive repository license and is in early-stage active development (v0.2.x release line).
Granular Permission System — Features an interactive selector to explicitly authorize or deny agent access to specific file paths and shell commands. - MCP Tool Integration — Supports Model Context Protocol (MCP) to connect external tool servers, complete with dedicated result rendering via MCPResultView. - Filesystem Tools — Includes built-in tools for workspace traversal, directory verification, and error-handled file inspection. - Terminal UI Components — Provides interactive command-line components for reviewing proposed changes, inspecting tool calls, and managing execution permissions. - Graceful Process Management — Automatically manages subprocess lifecycles, ensuring clean shutdown of spawned MCP servers and tools upon application exit.
Use cases
Use Case: Safe Workspace Exploration and File Inspection A developer working in an unfamiliar codebase can run mini-kode to locate relevant files and inspect implementations. The agent leverages its fileRead tooling while honoring path-level permission prompts before reading sensitive directories.
Use Case: Extensible Tooling with MCP Servers Teams that rely on external tools and databases can connect MCP servers directly to mini-kode. The CLI agent dispatches queries through the MCP interface and displays formatted tool results in the terminal for real-time verification.
Use Case: Interactive Command Approval and Code Editing When executing multi-step refactoring workflows, mini-kode presents each proposed shell command and file modification to the user via interactive terminal selection, preventing unauthorized file writes or dangerous script executions.
Who it is for
mini-kode is suited for developers and software engineers who want a lightweight, open-source terminal coding agent with explicit user consent controls and Model Context Protocol support. It is less suitable for users looking for turnkey graphical IDE integrations or fully autonomous agents running unattended in production CI pipelines.
Install guide
Installation Clone the repository and install dependencies using Node.js/npm (standard build method): bash git clone https://github.com/minmaxflow/mini-kode.git cd mini-kode npm install npm run build
First Run Ensure your LLM provider API keys (such as Anthropic or OpenAI) are exported in your environment before starting: bash export ANTHROPIC_API_KEY="your-api-key-here" node dist/index.js
Example Session Launch the agent in your project root and provide a task: bash # Start mini-kode in the current directory mini-kode # Example prompt inside the CLI session: > "Inspect src/index.ts and explain how MCP servers are initialized."
OpenCode — OpenCode is an open-source AI coding agent developed by Anomaly that operates across terminal environments, desktop applications, and IDE extensions.…
goose — goose is an open-source, general-purpose AI agent built in Rust, hosted under the vendor-neutral Agentic AI Foundation (AAIF) at the…
SWE-agent — SWE-agent is an open-source autonomous software engineering agent harness built and maintained by researchers from Princeton University and Stanford University.…
OpenHands — OpenHands (formerly OpenDevin) is an open-source autonomous AI software development agent and orchestration platform. It provides an extensible runtime environment…
Plandex — Plandex is an open-source terminal-based AI coding agent designed to handle complex, multi-file engineering tasks in large codebases. Created by…
Weaver — Weaver is an open-source agent harness and orchestration framework maintained by sean35mm on GitHub. Built on top of the Bun…