Indusagi Documentation
Choose the documentation you need
Indusagi Package
TypeScript SDK & Components
Coding Agent CLI
Terminal-First AI Assistant
NPM Packages
Real-time download statistics from npm registry
Frequently Asked Questions
Direct answers to common technical and general inquiries about Indusagi
What is Indusagi?
Indusagi is an open-source developer framework consisting of a TypeScript SDK and a terminal-first Coding Agent CLI. It is engineered to give developers granular programmatic control over autonomous AI agent loops, memory directories, and console UI rendering.
Do I have to pay to use Indusagi?
No. Both the TypeScript SDK and the CLI package are completely free to use under the MIT open-source license. You can deploy it for personal, institutional, or production-level commercial services without licensing fees.
Do I need cloud-based LLM APIs to run Indusagi?
No. While cloud APIs from Anthropic, OpenAI, and Groq are fully integrated, Indusagi natively supports local inference. You can configure it to execute completely offline using local providers such as Ollama and Llama.cpp.
How long does it take to set up and get started?
Initial environment initialization takes less than two minutes. You can install the CLI directly via npm, run the setup wizard to connect to your preferred model provider, and start working immediately.
What makes Indusagi different from other AI coding agents?
Unlike rigid, black-box AI platforms, Indusagi provides full code-level control. Developers can write custom hooks to intercept file operations, design dynamic console UI widgets, and configure granular token boundaries.
How does the CLI manage token usage and context windows?
The terminal CLI uses an automated session compaction algorithm. It periodically compiles extended conversation histories and active branch logs into semantic text summaries, preserving long-term project context while reducing API costs.
Do you support Model Context Protocol (MCP) servers?
Yes. The framework includes built-in client capabilities for MCP. This enables your custom agents to interact seamlessly with any third-party tools, local databases, or external microservices running the protocol.
Do I need to compile TypeScript files to build CLI extensions?
No. The Coding Agent CLI loads dynamic extension modules at runtime using the jiti compilation loader. You can write and execute TypeScript extension files immediately without manual pre-compilation.
How do I customize the terminal UI (TUI) widgets?
The SDK package includes a set of modular terminal widgets for dashboard design. You can programmatically set custom layouts, loading states, and text color schemes within your agent startup configuration.