Docs/TypeScript/Indusagi Coding Agent
StartREADME

Indusagi Coding Agent

A terminal-first AI coding agent. Point it at a repo and it reads, writes, runs, and reasons about your code. It runs entirely on the indusagi framework — one dependency for the LLM API, agent loop, terminal UI, MCP client, and tracing.

Built from scratch: every module was written from a behavioral spec and the framework's public API, never by copying third-party application source.

Install

npm install -g indusagi-coding-agent

This installs two bins that point at the same entry — use whichever you prefer:

indus       # start an interactive session in the current repo
indusagi    # identical bin

Requires Node >= 20.

Quick Start

indus                                       # interactive session in the current repo
indus "fix the failing test in src/auth"    # interactive, with an opening prompt
indus -p "summarise what changed on this branch"   # headless, prints result, exits
indus --json                                # newline-delimited JSON line protocol

Authenticate by exporting a provider key (e.g. ANTHROPIC_API_KEY) or run indus signin. In an interactive session you can also use /login. See Providers.

Modes

The entry resolves the invocation and selects one of three runners.

Invocation Mode What it does
indus interactive Ink/React REPL — streaming replies, a live tool deck, slash commands
indus -p "…" / --print print One request, prints the final result, exits
indus --json / --rpc line protocol Newline-delimited JSON for a parent process to drive
indus signin / signout credentials Store / clear a credential in the local auth vault

indus --help / -h prints the generated flag reference; indus --version / -v prints the version. --interactive / -i forces the REPL even when a prompt is supplied. See JSON Mode and RPC Mode for the headless protocols.

Key Features

  • Multi-provider support — Anthropic, OpenAI, Google, xAI, Groq, Cerebras, Mistral, OpenRouter, and more (see Providers).
  • Browser sign-in — Anthropic (Claude), OpenAI (ChatGPT), and GitHub Copilot can be authenticated through the browser via signin / /login.
  • Built-in tools — file read/write/edit, bash, grep, find, ls, web fetch, web search, task spawning, and todo tracking.
  • Session management — branch, resume, and continue prior sessions; automatic context-window compaction.
  • MCP — attach external MCP servers with --mcp or /mcp.
  • Extensible — load extensions, skills, prompt templates, and themes (Extensions, Skills, Themes).

Configuration paths

State lives under a single brand profile directory, ~/.indusagi/agent:

Path Holds
~/.indusagi/agent/settings.json Global settings (all projects)
.indusagi/settings.json Project settings (resolved against the working directory)
~/.indusagi/agent/auth.json Credential vault (api keys and browser-sign-in tokens)
~/.indusagi/agent/sessions/ Saved sessions

Set INDUSAGI_CODING_AGENT_DIR to relocate the entire profile directory. See Settings.

Documentation Map

Getting Started

  • Architecture - Module map and the entry → boot → runner flow
  • Providers - Configure LLM providers and sign in
  • Settings - Global and project settings

Modes & API

Customization

Session Management

Configuration

UI

License

MIT

Next →
Architecture: Module Map