stynx-code

An interactive AI coding assistant for the terminal — multi-provider, tool-using, fast, and yours.

  ____ _____ __   ___   __ __
 / ___|_   _\ \ / / \ | \ \ / /
 \___ \ | |  \ V /|  \| |\ V /
  ___) || |   | | | |\  | | |
 |____/ |_|   |_| |_| \_| |_|
               c o d e

What it is

stynx-code is a self-contained TUI for getting work done with an LLM. It speaks Anthropic's Claude API natively and any OpenAI-compatible endpoint (DeepSeek, OpenAI, OpenRouter, Together, local Ollama, …) for cheap delegation. It runs an autonomous tool-using loop with bash / file edit / glob / grep / web fetch and a permission system that prompts you only when you want it to.

Highlights:

• Modern terminal UI — sidebar, command palette, model picker, session list, theme switcher (rose-pine, catppuccin, tokyo-night, gruvbox), mouse + bracketed paste, @-file mentions, multi-line input, slash-command popover, toast notifications, diff renderer for file edits.
• Intern mode — Claude as the senior, DeepSeek (or any OpenAI-compat model) as the intern. Senior delegates focused subtasks via the delegate_to_intern tool or /intern <task> slash command.
• Permission-first — Normal / Auto-accept / Plan / Bypass modes. Per-tool allow rules. Inline confirmation modal — no terminal mode-switching.
• First-class skills — drop a markdown file under .claude/skills/ and it shows up as a slash command.
• Sessions — automatic per-project persistence at ~/.stynx-code/projects/<slug>/session.json.
• Hooks — session-start, pre-tool-use, post-tool-use shell hooks for integrations.

Install

# from source, with cargo
cargo install --path stynx-code

# or with nix
nix develop
cargo build --release --workspace

The binary is stynx-code.

Quickstart

# interactive TUI (the main mode)
stynx-code

# one-shot prompt
stynx-code -p "find every TODO comment under src/ and group by file"

# pipe mode — read stdin as the prompt
echo "explain this file" | stynx-code

# JSON output for scripting
stynx-code --json -p "list files modified in the last commit"

Configuration

Anthropic credentials — uses Claude Code's OAuth token (~/.claude/) if present, falls back to ANTHROPIC_API_KEY. Run /login for instructions.

Intern (OpenAI-compatible) provider — set in .env at the project root or as shell env:

DEEPSEEK_API_KEY=sk-...
# optional overrides:
DEEPSEEK_MODEL=deepseek-chat                  # default
DEEPSEEK_BASE_URL=https://api.deepseek.com/v1

Add .env to your .gitignore.

Settings file — ~/.config/stynx-code/settings.json (or project-local .stynx/settings.json) for permission rules, model defaults, hooks. Run /config to see the merged view.

Daily flow

Mouse scroll wheel works in any terminal with mouse capture.

Slash commands

/help, /status, /version, /quit, /exit /model [name], /fast, /effort low|medium|high|max, /think /mode, /plan [task] /compact, /cost, /usage /diff, /status, /review, /commit /memory, /init, /add <path>, /files, /skills /session, /rewind [n], /export, /copy, /undo [n] /config, /permissions /intern <task> — hand work to the intern model directly !<command> — run a shell command without leaving the TUI

Architecture

Every crate follows Clean Architecture with domain/, application/, and infrastructure/ layers.

stynx-code/                main binary, app loop, CLI surface
stynx-code-types/          shared types: Provider, Tool, Message, Conversation
stynx-code-errors/         AppError + AppResult
stynx-code-config/         settings loader, theme/keybind config
stynx-code-auth/           credential resolution (Anthropic OAuth / env keys)
stynx-code-provider/       AnthropicProvider + OpenAiProvider (DeepSeek etc.)
stynx-code-engine/         streaming tool-use loop, max-turns, compact
stynx-code-tools/          bash, read, write, edit, glob, grep, web_fetch, …
stynx-code-permission/     allow/deny rules + interactive prompts (bridge to TUI)
stynx-code-memory/         per-project session persistence
stynx-code-commands/       slash-command handlers
stynx-code-compact/        conversation summarization
stynx-code-coordinator/    parallel sub-agent orchestration
stynx-code-skills/         user-defined skill loading
stynx-code-plugins/        plugin host (skills + MCP)
stynx-code-bridge/         server bridge for IDE/web frontends
stynx-code-server/         optional HTTP/SSE server surface
stynx-code-services/       cross-cutting services (tips, telemetry, …)
stynx-code-tui/            ratatui-based terminal UI

Intern mode in 30 seconds

# .env
DEEPSEEK_API_KEY=sk-...

# in stynx-code
> refactor every println! in src/foo.rs to tracing::info!; hand it to the intern

Claude calls delegate_to_intern with a focused task description. The intern (DeepSeek) runs with a restricted toolset (bash/read/write/edit/glob/grep — no recursive sub-agents) and returns Summary / Files changed / Output. Claude reviews and integrates.

You can also call it directly: /intern list every public function in stynx-code-tui.

Hooks

Drop scripts referenced by settings.json::hooks:

• session-start — run when a session begins; output is shown to the user
• pre-tool-use — run before each tool execution; can inject context
• post-tool-use — run after each tool execution; can post-process

Each hook receives the relevant JSON on stdin and prints text to stdout.

License

MIT