Login

Changelog

New updates and improvements to supercode

v0.1.7

Jun 19, 2026

CLI / TUI

  • - Phosphor CRT terminal theme — green phosphor on black, amber accent for cursor/badges, green glow highlights. New pixelWordmark() renders "SUPERCODE" as ASCII pixel art.
  • - Main screen redesigned with sectionHeader, cardStack, rowCard, keyValue, statusBar utilities — centered pixel wordmark, system HUD panel, command card stack, and a footer status bar.
  • - Chat startup overhauled: pixel wordmark header, status bar with mode/model info, /model / /help / Tab quick-start hint row.
  • - Framed "goodbye" message on exit instead of a single line.
  • - Animated thinking spinner with live tool call and reasoning updates (existing, refined).
  • - 1,048 insertions across 7 files — the TUI is now the primary surface identity.

Docs

  • - README rewritten with new branding ("AI-Powered SWE Agent"), updated app/package table, Terminal Stack architecture section, and new badges (Next.js 16, License).
  • - CONTRIBUTING rewritten from scratch: Bun-first setup, both database setup steps (dashboard + terminal CLI), expanded project tree, full command reference, code style guidelines, CLI pattern doc, DB migration workflows.
  • - Config documentation updated with production URL guidance for OAuth setup — thanks @aviisharma238!

v0.1.6

Jun 11, 2026

Core

  • - Agent mode rewritten from rigid generateApplication() to tool-calling loop (ToolLoopAgent) — model calls write_file and run_command tools directly for iterative building and error recovery.
  • - New /model slash command — switch AI provider/model mid-session (Gemini, OpenRouter, NVIDIA NIM).
  • - Persistent stdin handler — prevents "terminal state may be corrupted" crashes.
  • - Stream cancellation via Escape key with partial response preservation.
  • - Global unhandledRejection and uncaughtException handlers for crash resilience.

Tools

  • - New standalone tool definitions: write-file, run-command, read-instructions with path traversal protection, 1MB size limit, and auto mkdir support.
  • - PermissionManager — granular tool permission system with wildcard matching, dangerous command detection (20 regex patterns), and [y] Once / [a] Always / [n] Deny prompts.
  • - OpenRouter provider rewritten from SDK to raw fetch API with native tool call detection and multi-iteration tool loops.
  • - NVIDIA NIM provider migrated to @ai-sdk/openai-compatible SDK with proper tool-calling support.

CLI

  • - Model switching at runtime via /model command supporting Google Gemini, OpenRouter (GPT OSS, DeepSeek V4 Flash, MiniMax M3, GLM 5.1, Kimi K2.6), and NVIDIA NIM (MiniMax M2.7, DeepSeek V4 Flash, Llama 3.3 70B).
  • - Default OpenRouter model changed to moonshotai/kimi-k2.6:free.
  • - Animated thinking spinner with live tool call and reasoning updates.
  • - Agent mode shows live tool call progress with exit and continue prompts.
  • - Chat loop resilience — catch-all error handling prevents crashes; terminal state restored on errors.

v0.1.5

Jun 8, 2026

Core

  • - New OpenRouter models: MiniMax M3, GLM 5.1, and Kimi K2.6 (free).
  • - Session token tracking with context window percentage display (136.4K (68%)) in chat footer.
  • - maxOutputTokens cap (8192) for paid OpenRouter models to prevent 402 errors.
  • - Render cold start error message improved — "server was inactive" instead of "cannot reach server".
  • - Human-readable token formatting (5.2K, 136.4K, 1.0M).

Docs

  • - Rewamp authentication docs covering CLI device code flow, token storage, GitHub OAuth, and Render cold start troubleshooting.
  • - Rewamp terminal architecture docs reflecting actual codebase (CLI commands, DB models, API endpoints, AI providers, deployment).

v0.1.4

Jun 8, 2026

Core

  • - Server-side AI streaming endpoint (/api/ai/chat) for proxied chat completions through Google, OpenRouter, MiniMax, and NVIDIA.
  • - ServerProxyService for secure AI provider request routing (API keys stay server-side).
  • - Structured generation endpoint (/api/ai/generate-object) for schema-guided AI output.
  • - CLI version display banner showing engine, auth, and model configuration.
  • - NVIDIA NIM streaming support with token usage tracking.

CLI

  • - supercode init added as package script.
  • - supercode login added as package script.
  • - CLI commands executable via bun run in server package.

Web App

  • - Metadata updates and favicon added to dashboard.

v0.1.3

Jun 8, 2026

Core

  • - Clearer AI provider configuration errors — each service now shows which env var to set when a key is missing.
  • - Provider logic refactored with per-service env var hints for Google, MiniMax, OpenRouter, and NVIDIA.
  • - Enhanced error messages for API key configuration.

v0.1.2

Jun 7, 2026

Core

  • - AI service integration refactor with improved error handling and model configuration.
  • - API client timeout support (15s default) for reliable network requests.
  • - OpenRouter AI SDK provider integrated alongside existing providers.
  • - Bearer token validation via Prisma session lookup for API auth.
  • - Device authorization form component for secure device-login UX.
  • - CORS configuration improved with server URL support.

CLI

  • - Environment-aware config — loads .env from working directory, package path, and dist directory.
  • - Tool-assisted chat mode with interactive tool selection (multiselect).
  • - AI agent mode for generating full applications from descriptions.
  • - Conversation mode switching (chat / tools / agent).
  • - Workspace scanning with file tree rendering.
  • - Model selection prompts with provider-specific model lists.
  • - Documentation URL corrected in CLI banner.

Web App

  • - Device authorization page with code input form.
  • - Sign-in page with GitHub OAuth and email/password via Better-Auth.
  • - Redirect and error handling routes for auth flows.
  • - Next.js turbopack support with root path resolution.

v0.1.0

Jun 6, 2026

Core

  • - AI coding agent with multi-provider support — Google Gemini, OpenAI, Anthropic Claude, Minimax, NVIDIA, and OpenRouter.
  • - Free models included — GPT, Gemini, Minimax, Deepseek and more available out of the box.
  • - AI-powered chat interface with streaming responses, tool execution, and workspace context awareness.
  • - Device code authentication flow with GitHub OAuth for secure CLI login.
  • - WebSocket server for real-time AI streaming and tool execution.
  • - Multi-session support with conversation history and message persistence.
  • - Prisma-backed database with PostgreSQL for users, sessions, workspaces, and API keys.

CLI

  • - New supercode init command with interactive chat and agent modes.
  • - supercode login command with device code authentication.
  • - Environment-aware configuration with .env loading from multiple locations.
  • - Tool system for AI — read files, search code, list directories, and execute commands.

Web App

  • - Terminal client at supercode-terminal.vercel.app with session management and workspace file explorer.
  • - Device authorization and approval pages for secure device login.
  • - Sign-in page with email/password and GitHub OAuth via Better-Auth.