Home / Features

Features

A complete AI coding toolkit. Works in your terminal and your browser.

3 FREE AI Providers

Gemini, Groq, and OpenRouter — all with generous free tiers. Switch between them anytime with a single command. No credit card needed.

Streaming Responses

See AI responses appear word-by-word in real-time. Both the CLI and Web Chat stream output as it's generated — zero waiting.

Shared Config & API Keys

Set your API keys once. Both CLI and Web Chat read the same config.json. No duplicate setup, no hassle.

File Operations

Read, write, edit, and delete files directly through conversation. The AI can understand your project structure and make precise changes.

Command Execution

Run terminal commands and see output in real-time. The AI can install packages, run tests, build projects, and deploy.

Interactive Permissions

Control what the AI can do automatically. Choose from 4 modes: default, auto-edit, plan, or yolo — with granular tool-level control.

Session Management

Save, resume, and manage chat sessions on both interfaces. Each session preserves context, model settings, and conversation history.

Code Search

Find files by glob patterns with integrated code search. The AI navigates your entire project structure to find what you need.

Token Tracking

Monitor usage with /tokens — see exact counts per session. Stay informed about your API consumption in real-time.

Autocomplete Popup

Type / to see available commands. Context-aware suggestions help you discover features without memorizing syntax.

Web Chat — Instant

PKA CODE works in your browser via pkacodeweb. Chat instantly with DeepSeek V4 Flash — no API key needed.

</> Browser-Based Chat

Run pkacodeweb to start the local server and open http://localhost:3721/chat in any browser. Chat instantly with DeepSeek V4 Flash — no API key, no signup required.

Real-Time Streaming

Every AI response streams word-by-word via Server-Sent Events (SSE). See the answer form in real-time, with a blinking cursor indicating live generation.

Switch Providers On-the-Fly

Change between Gemini, Groq, and OpenRouter with a simple dropdown. Models update automatically based on your selected provider.

Settings Panel

Set API keys, custom system prompts, adjust temperature, and configure max tokens — all from a clean settings modal.

Session Saving

Conversations are auto-saved and listed in the sidebar. Resume any previous chat, or start fresh with /new.

Mobile-Responsive

Full touch support with swipe gestures, a bottom navigation bar, and a slide-out drawer sidebar. Works on phones, tablets, and desktops.

Web Chat Setup Guide →

Permission modes

Granular control over what the AI can do automatically. Choose the mode that fits your workflow.

Default
Read-only auto, edits and bash ask for your confirmation. Safe and balanced.
Auto-edit
Read/write/edit auto, bash asks you. For when you trust the AI with file changes.
Plan
Read-only only — no edits, no bash. Perfect for code review and analysis.
Yolo
Everything automatic — no prompts. Maximum speed, minimum safety rails.

Supported providers

All free. All production-ready. Switch providers anytime without losing your session.

Provider Free Tier Free Models Get Key
Gemini 60 req/min 5 models (Flash, Flash-Lite, etc.) Google AI Studio →
Groq 30 req/min 6 models (Llama, Mixtral, Gemma, Qwen) Groq Console →
OpenRouter 20+ free models DeepSeek, Qwen, Phi, Cohere, and more OpenRouter →

Cross-platform

Works on every major operating system. Designed for developers, regardless of their platform choice.

Platform CLI Status Web Chat Status Config Path
Windows 10+ Supported Supported %APPDATA%\pka\config.json
Linux Supported Supported ~/.config/pka/config.json
macOS Supported Supported ~/.config/pka/config.json

Available free models

21 free models across 3 providers. The best AI models money can't buy.

Gemini

  • gemini-2.0-flash — Best free default, vision
  • gemini-2.5-flash — Newer, smarter, 1M context
  • gemini-2.5-flash-lite — Lowest latency, highest quota
  • gemini-3.1-flash-lite — Newest, most cost-efficient
  • gemini-2.0-flash-lite — Lightweight, fast responses

Groq

  • llama-3.3-70b-versatile — Strong general purpose
  • llama-3.1-70b-versatile — Older but capable
  • llama-3.1-8b-instant — Fastest, lightweight
  • mixtral-8x7b-32768 — MoE, strong reasoning
  • gemma2-9b-it — Google, balanced and fast
  • qwen/qwen3-32b — Coding and math focused

OpenRouter (10+ free models)

  • openrouter/free — Auto router, best free model
  • google/gemini-2.0-flash-exp:free — Fast, vision
  • google/gemini-2.5-flash-exp:free — Newer Gemini
  • meta-llama/llama-3.3-70b-instruct:free — Strong general
  • deepseek/deepseek-chat:free — Strong reasoning
  • qwen/qwen-2.5-coder-32b-instruct:free — Best for coding
  • microsoft/phi-3-medium:free — Compact and capable
  • …and more auto-discovered via API