claudish 3.1.1 → 3.1.3

package/AI_AGENT_GUIDE.md

@@ -1,6 +1,6 @@
 # Claudish AI Agent Usage Guide
 
-**Version:** 1.0.0
+**Version:** 2.0.0
 **Target Audience:** AI Agents running within Claude Code
 **Purpose:** Quick reference for using Claudish CLI in agentic workflows
 
@@ -10,24 +10,43 @@
 
 ```bash
 # 1. Get available models
-claudish --list-models --json
+claudish --models --json
+
+# 2. Run task with specific model (OpenRouter)
+claudish --model openai/gpt-5.2 "your task here"
+
+# 3. Run with direct Gemini API
+claudish --model g/gemini-2.0-flash "your task here"
 
-# 2. Run task with specific model
-claudish --model x-ai/grok-code-fast-1 "your task here"
+# 4. Run with local model
+claudish --model ollama/llama3.2 "your task here"
 
-# 3. For large prompts, use stdin
-echo "your task" | claudish --stdin --model x-ai/grok-code-fast-1
+# 5. For large prompts, use stdin
+echo "your task" | claudish --stdin --model openai/gpt-5.2
 ```
 
 ## What is Claudish?
 
-Claudish = Claude Code + OpenRouter models
+Claudish = Claude Code + Any AI Model
 
-- ✅ Run Claude Code with **any OpenRouter model** (Grok, GPT-5, Gemini, MiniMax, etc.)
+- ✅ Run Claude Code with **any AI model** via prefix-based routing
+- ✅ Supports OpenRouter (100+ models), direct Gemini API, direct OpenAI API
+- ✅ Supports local models (Ollama, LM Studio, vLLM, MLX)
 - ✅ 100% Claude Code feature compatibility
 - ✅ Local proxy server (no data sent to Claudish servers)
 - ✅ Cost tracking and model selection
 
+## Model Routing
+
+| Prefix | Backend | Example |
+|--------|---------|---------|
+| _(none)_ | OpenRouter | `openai/gpt-5.2` |
+| `g/` `gemini/` | Google Gemini | `g/gemini-2.0-flash` |
+| `oai/` `openai/` | OpenAI | `oai/gpt-4o` |
+| `ollama/` | Ollama | `ollama/llama3.2` |
+| `lmstudio/` | LM Studio | `lmstudio/model` |
+| `http://...` | Custom | `http://localhost:8000/model` |
+
 ## Prerequisites
 
 1. **Install Claudish:**
@@ -49,15 +68,25 @@ Claudish = Claude Code + OpenRouter models
 
 | Model ID | Provider | Category | Best For |
 |----------|----------|----------|----------|
-| `x-ai/grok-code-fast-1` | xAI | Coding | Fast iterations, agentic coding |
-| `google/gemini-2.5-flash` | Google | Reasoning | Complex analysis, 1000K context |
-| `minimax/minimax-m2` | MiniMax | Coding | General coding tasks |
-| `openai/gpt-5` | OpenAI | Reasoning | Architecture decisions |
-| `qwen/qwen3-vl-235b-a22b-instruct` | Alibaba | Vision | UI/visual tasks |
+| `openai/gpt-5.2` | OpenAI | Reasoning | **Default** - Most advanced reasoning |
+| `minimax/minimax-m2.1` | MiniMax | Coding | Budget-friendly, fast |
+| `z-ai/glm-4.7` | Z.AI | Coding | Balanced performance |
+| `google/gemini-3-pro-preview` | Google | Reasoning | 1M context window |
+| `moonshotai/kimi-k2-thinking` | MoonShot | Reasoning | Extended thinking |
+| `deepseek/deepseek-v3.2` | DeepSeek | Coding | Code specialist |
+| `qwen/qwen3-vl-235b-a22b-thinking` | Alibaba | Vision | Vision + reasoning |
+
+**Direct API Options (lower latency):**
+
+| Model ID | Backend | Best For |
+|----------|---------|----------|
+| `g/gemini-2.0-flash` | Gemini | Fast tasks, large context |
+| `oai/gpt-4o` | OpenAI | General purpose |
+| `ollama/llama3.2` | Local | Free, private |
 
 **Update models:**
 ```bash
-claudish --list-models --force-update
+claudish --models --force-update
 ```
 
 ## Critical: File-Based Pattern for Sub-Agents
@@ -529,6 +558,6 @@ claudish --help-ai > claudish-agent-guide.md
 
 ---
 
-**Version:** 1.0.0
-**Last Updated:** November 19, 2025
+**Version:** 2.0.0
+**Last Updated:** January 5, 2026
 **Maintained by:** MadAppGang

package/README.md

@@ -1,17 +1,19 @@
 # Claudish
 
-> Run Claude Code with OpenRouter models via local proxy
+> Run Claude Code with any AI model - OpenRouter, Gemini, OpenAI, or local models
 
-**Claudish** (Claude-ish) is a CLI tool that allows you to run Claude Code with any OpenRouter model by proxying requests through a local Anthropic API-compatible server.
+**Claudish** (Claude-ish) is a CLI tool that allows you to run Claude Code with any AI model by proxying requests through a local Anthropic API-compatible server. Supports OpenRouter (100+ models), direct Google Gemini API, direct OpenAI API, and local models (Ollama, LM Studio, vLLM, MLX).
 
 ## Features
 
+- ✅ **Multi-provider support** - OpenRouter, Gemini, OpenAI, and local models via prefix routing
+- ✅ **Direct API access** - Use `g/gemini-2.0-flash` or `oai/gpt-4o` for direct API calls
+- ✅ **Local model support** - Ollama, LM Studio, vLLM, MLX with `ollama/`, `lmstudio/` prefixes
 - ✅ **Cross-platform** - Works with both Node.js and Bun (v1.3.0+)
 - ✅ **Universal compatibility** - Use with `npx` or `bunx` - no installation required
 - ✅ **Interactive setup** - Prompts for API key and model if not provided (zero config!)
 - ✅ **Monitor mode** - Proxy to real Anthropic API and log all traffic (for debugging)
 - ✅ **Protocol compliance** - 1:1 compatibility with Claude Code communication protocol
-- ✅ **Snapshot testing** - Comprehensive test suite with 13/13 passing tests
 - ✅ **Headless mode** - Automatic print mode for non-interactive execution
 - ✅ **Quiet mode** - Clean output by default (no log pollution)
 - ✅ **JSON output** - Structured data for tool integration
@@ -19,7 +21,6 @@
 - ✅ **Parallel runs** - Each instance gets isolated proxy
 - ✅ **Autonomous mode** - Bypass all prompts with flags
 - ✅ **Context inheritance** - Runs in current directory with same `.claude` settings
-- ✅ **Multiple models** - 10+ prioritized OpenRouter models
 - ✅ **Agent support** - Use Claude Code agents in headless mode with `--agent`
 
 ## Installation
@@ -43,7 +44,11 @@ bun install -g claudish
 ### Prerequisites
 
 - [Claude Code](https://claude.com/claude-code) - Claude CLI must be installed
-- [OpenRouter API Key](https://openrouter.ai/keys) - Free tier available
+- At least one API key:
+  - [OpenRouter API Key](https://openrouter.ai/keys) - Access 100+ models (free tier available)
+  - [Google Gemini API Key](https://aistudio.google.com/apikey) - For direct Gemini access
+  - [OpenAI API Key](https://platform.openai.com/api-keys) - For direct OpenAI access
+  - Or local models (Ollama, LM Studio) - No API key needed
 
 ### Other Install Options
 
@@ -228,46 +233,94 @@ claudish [OPTIONS] <claude-args...>
 
 ### Environment Variables
 
-| Variable | Description | Required |
+#### API Keys (at least one required)
+
+| Variable | Description | Used For |
 |----------|-------------|----------|
-| `OPENROUTER_API_KEY` | Your OpenRouter API key | ⚡ **Optional in interactive mode** (will prompt if not set)<br>✅ **Required in non-interactive mode** |
-| `ANTHROPIC_API_KEY` | Placeholder to prevent Claude Code dialog (not used for auth) | ✅ **Required** |
-| `CLAUDISH_MODEL` | Default model to use | ❌ No |
-| `CLAUDISH_PORT` | Default proxy port | ❌ No |
-| `CLAUDISH_ACTIVE_MODEL_NAME` | Automatically set by claudish to show active model in status line (read-only) | ❌ No |
+| `OPENROUTER_API_KEY` | OpenRouter API key | Default backend (100+ models) |
+| `GEMINI_API_KEY` | Google Gemini API key | Direct Gemini access (`g/` prefix) |
+| `OPENAI_API_KEY` | OpenAI API key | Direct OpenAI access (`oai/` prefix) |
+| `ANTHROPIC_API_KEY` | Placeholder (any value) | Prevents Claude Code dialog |
+
+#### Custom Endpoints (optional)
+
+| Variable | Description | Default |
+|----------|-------------|---------|
+| `GEMINI_BASE_URL` | Custom Gemini endpoint | `https://generativelanguage.googleapis.com` |
+| `OPENAI_BASE_URL` | Custom OpenAI/Azure endpoint | `https://api.openai.com` |
+| `OLLAMA_BASE_URL` | Ollama server URL | `http://localhost:11434` |
+| `LMSTUDIO_BASE_URL` | LM Studio server URL | `http://localhost:1234` |
+| `VLLM_BASE_URL` | vLLM server URL | `http://localhost:8000` |
+| `MLX_BASE_URL` | MLX server URL | `http://127.0.0.1:8080` |
+
+#### Other Settings
+
+| Variable | Description | Default |
+|----------|-------------|---------|
+| `CLAUDISH_MODEL` | Default model to use | `openai/gpt-5.2` |
+| `CLAUDISH_PORT` | Default proxy port | Random (3000-9000) |
+| `CLAUDISH_CONTEXT_WINDOW` | Override context window size | Auto-detected |
 
 **Important Notes:**
-- **NEW in v1.3.0:** In interactive mode, if `OPENROUTER_API_KEY` is not set, you'll be prompted to enter it
 - You MUST set `ANTHROPIC_API_KEY=sk-ant-api03-placeholder` (or any value). Without it, Claude Code will show a dialog
+- In interactive mode, if no API key is set, you'll be prompted to enter one
+
+## Model Routing (v3.1.0+)
+
+Claudish uses **prefix-based routing** to determine which API backend to use:
+
+| Prefix | Backend | API Key | Example |
+|--------|---------|---------|---------|
+| _(none)_ | OpenRouter | `OPENROUTER_API_KEY` | `openai/gpt-5.2` |
+| `or/` | OpenRouter | `OPENROUTER_API_KEY` | `or/anthropic/claude-3.5-sonnet` |
+| `g/` `gemini/` `google/` | Google Gemini | `GEMINI_API_KEY` | `g/gemini-2.0-flash` |
+| `oai/` `openai/` | OpenAI | `OPENAI_API_KEY` | `oai/gpt-4o` |
+| `ollama/` | Ollama | _(none)_ | `ollama/llama3.2` |
+| `lmstudio/` | LM Studio | _(none)_ | `lmstudio/qwen2.5-coder` |
+| `vllm/` | vLLM | _(none)_ | `vllm/mistral-7b` |
+| `mlx/` | MLX | _(none)_ | `mlx/llama-3.2-3b` |
+| `http://...` | Custom | _(none)_ | `http://localhost:8000/model` |
 
-## Available Models
+### Examples
 
-Claudish supports 5 OpenRouter models in priority order:
+```bash
+# OpenRouter (default) - 100+ models via unified API
+claudish --model openai/gpt-5.2 "implement feature"
+claudish --model anthropic/claude-3.5-sonnet "review code"
+
+# Direct Gemini API - lower latency, direct billing
+claudish --model g/gemini-2.0-flash "quick task"
+claudish --model gemini/gemini-2.5-pro "complex analysis"
 
-1. **x-ai/grok-code-fast-1** (Default)
-   - Fast coding-focused model from xAI
-   - Best for quick iterations
+# Direct OpenAI API - lower latency, direct billing
+claudish --model oai/gpt-4o "implement feature"
+claudish --model openai/o1 "complex reasoning"
 
-2. **openai/gpt-5-codex**
-   - Advanced coding model from OpenAI
-   - Best for complex implementations
+# Local models - free, private, no API key needed
+claudish --model ollama/llama3.2 "code review"
+claudish --model lmstudio/qwen2.5-coder "refactor"
+```
 
-3. **minimax/minimax-m2**
-   - High-performance model from MiniMax
-   - Good for general coding tasks
+## Curated Models
 
-4. **zhipu-ai/glm-4.6**
-   - Advanced model from Zhipu AI
-   - Good for multilingual code
+Top recommended models for development (v3.1.1):
 
-5. **qwen/qwen3-vl-235b-a22b-instruct**
-   - Vision-language model from Alibaba
-   - Best for UI/visual tasks
+| Model | Provider | Best For |
+|-------|----------|----------|
+| `openai/gpt-5.2` | OpenAI | **Default** - Most advanced reasoning |
+| `minimax/minimax-m2.1` | MiniMax | Budget-friendly, fast |
+| `z-ai/glm-4.7` | Z.AI | Balanced performance |
+| `google/gemini-3-pro-preview` | Google | 1M context window |
+| `moonshotai/kimi-k2-thinking` | MoonShot | Extended reasoning |
+| `deepseek/deepseek-v3.2` | DeepSeek | Code specialist |
+| `qwen/qwen3-vl-235b-a22b-thinking` | Alibaba | Vision + reasoning |
 
-List models anytime with:
+List all models:
 
 ```bash
-claudish --models
+claudish --models              # List all OpenRouter models
+claudish --models gemini       # Search for specific models
+claudish --top-models          # Show curated recommendations
 ```
 
 ## Agent Support (NEW in v2.1.0)
@@ -363,14 +416,19 @@ claudish "implement user authentication with JWT tokens"
 ### With Specific Model
 
 ```bash
-# Use Grok for fast coding
-claudish --model x-ai/grok-code-fast-1 "add error handling"
+# Use OpenRouter models (default)
+claudish --model openai/gpt-5.2 "refactor entire API layer"
+claudish --model deepseek/deepseek-v3.2 "add error handling"
+
+# Use direct Gemini API (faster, direct billing)
+claudish --model g/gemini-2.0-flash "quick fix"
 
-# Use GPT-5 Codex for complex tasks
-claudish --model openai/gpt-5-codex "refactor entire API layer"
+# Use direct OpenAI API
+claudish --model oai/gpt-4o "implement feature"
 
-# Use Qwen for UI tasks
-claudish --model qwen/qwen3-vl-235b-a22b-instruct "implement dashboard UI"
+# Use local models (free, private)
+claudish --model ollama/llama3.2 "code review"
+claudish --model lmstudio/qwen2.5-coder "implement dashboard UI"
 ```
 
 ### Autonomous Mode

package/dist/index.js

@@ -35367,12 +35367,22 @@ function printVersion() {
 }
 function printHelp() {
   console.log(`
-claudish - Run Claude Code with OpenRouter models
+claudish - Run Claude Code with any AI model (OpenRouter, Gemini, OpenAI, Local)
 
 USAGE:
   claudish                                # Interactive mode (default, shows model selector)
   claudish [OPTIONS] <claude-args...>     # Single-shot mode (requires --model)
 
+MODEL ROUTING (prefix-based):
+  (no prefix)      OpenRouter (default)   claudish --model openai/gpt-5.2 "task"
+  g/, gemini/      Google Gemini API      claudish --model g/gemini-2.0-flash "task"
+  oai/, openai/    OpenAI API             claudish --model oai/gpt-4o "task"
+  ollama/          Ollama (local)         claudish --model ollama/llama3.2 "task"
+  lmstudio/        LM Studio (local)      claudish --model lmstudio/qwen "task"
+  vllm/            vLLM (local)           claudish --model vllm/model "task"
+  mlx/             MLX (local)            claudish --model mlx/model "task"
+  http://...       Custom endpoint        claudish --model http://localhost:8000/model "task"
+
 OPTIONS:
   -i, --interactive        Run in interactive mode (default when no prompt given)
   -m, --model <model>      OpenRouter model to use (required for single-shot mode)
@@ -35433,27 +35443,33 @@ NOTES:
 ENVIRONMENT VARIABLES:
   Claudish automatically loads .env file from current directory.
 
-  OPENROUTER_API_KEY              Required: Your OpenRouter API key (for OpenRouter models)
-  CLAUDISH_MODEL                  Default model to use (takes priority)
-  ANTHROPIC_MODEL                 Claude Code standard: model to use (fallback)
+  API Keys (at least one required for cloud models):
+  OPENROUTER_API_KEY              OpenRouter API key (default backend)
+  GEMINI_API_KEY                  Google Gemini API key (for g/ prefix)
+  OPENAI_API_KEY                  OpenAI API key (for oai/ prefix)
+  ANTHROPIC_API_KEY               Placeholder (prevents Claude Code dialog)
+
+  Custom endpoints:
+  GEMINI_BASE_URL                 Custom Gemini endpoint
+  OPENAI_BASE_URL                 Custom OpenAI/Azure endpoint
+
+  Local providers:
+  OLLAMA_BASE_URL                 Ollama server (default: http://localhost:11434)
+  OLLAMA_HOST                     Alias for OLLAMA_BASE_URL
+  LMSTUDIO_BASE_URL               LM Studio server (default: http://localhost:1234)
+  VLLM_BASE_URL                   vLLM server (default: http://localhost:8000)
+  MLX_BASE_URL                    MLX server (default: http://127.0.0.1:8080)
+
+  Model settings:
+  CLAUDISH_MODEL                  Default model to use (default: openai/gpt-5.2)
   CLAUDISH_PORT                   Default port for proxy
-  CLAUDISH_ACTIVE_MODEL_NAME      Auto-set by claudish (read-only) - shows active model
+  CLAUDISH_CONTEXT_WINDOW         Override context window size
 
-  Model mapping (CLAUDISH_* takes priority over ANTHROPIC_DEFAULT_*):
+  Model mapping (per-role):
   CLAUDISH_MODEL_OPUS             Override model for Opus role
   CLAUDISH_MODEL_SONNET           Override model for Sonnet role
   CLAUDISH_MODEL_HAIKU            Override model for Haiku role
   CLAUDISH_MODEL_SUBAGENT         Override model for sub-agents
-  ANTHROPIC_DEFAULT_OPUS_MODEL    Claude Code standard: Opus model (fallback)
-  ANTHROPIC_DEFAULT_SONNET_MODEL  Claude Code standard: Sonnet model (fallback)
-  ANTHROPIC_DEFAULT_HAIKU_MODEL   Claude Code standard: Haiku model (fallback)
-  CLAUDE_CODE_SUBAGENT_MODEL      Claude Code standard: sub-agent model (fallback)
-
-  Local providers (OpenAI-compatible):
-  OLLAMA_BASE_URL                 Ollama server (default: http://localhost:11434)
-  OLLAMA_HOST                     Alias for OLLAMA_BASE_URL (same default)
-  LMSTUDIO_BASE_URL               LM Studio server (default: http://localhost:1234)
-  VLLM_BASE_URL                   vLLM server (default: http://localhost:8000)
 
 EXAMPLES:
   # Interactive mode (default) - shows model selector
@@ -35463,26 +35479,28 @@ EXAMPLES:
   # Interactive mode with only FREE models
   claudish --free
 
-  # Interactive mode with pre-selected model
-  claudish --model x-ai/grok-code-fast-1
+  # OpenRouter models (default)
+  claudish --model openai/gpt-5.2 "implement user authentication"
+  claudish --model deepseek/deepseek-v3.2 "add tests for login"
 
-  # Single-shot mode - one task and exit (requires --model or CLAUDISH_MODEL env var)
-  claudish --model openai/gpt-5-codex "implement user authentication"
-  claudish --model x-ai/grok-code-fast-1 "add tests for login"
+  # Direct Gemini API (lower latency)
+  claudish --model g/gemini-2.0-flash "quick fix"
+  claudish --model gemini/gemini-2.5-pro "complex analysis"
 
-  # Per-role model mapping (use different models for different Claude Code roles)
-  claudish --model-opus openai/gpt-5 --model-sonnet x-ai/grok-code-fast-1 --model-haiku minimax/minimax-m2
+  # Direct OpenAI API
+  claudish --model oai/gpt-4o "implement feature"
+  claudish --model openai/o1 "complex reasoning"
 
-  # Use named profiles for pre-configured model mappings
-  claudish -p frontend "implement component"
-  claudish --profile debug "investigate error"
+  # Local models (free, private)
+  claudish --model ollama/llama3.2 "code review"
+  claudish --model lmstudio/qwen2.5-coder "refactor"
 
-  # Hybrid: Native Anthropic for Opus, OpenRouter for Sonnet/Haiku
-  claudish --model-opus claude-3-opus-20240229 --model-sonnet x-ai/grok-code-fast-1
+  # Per-role model mapping
+  claudish --model-opus openai/gpt-5.2 --model-sonnet deepseek/deepseek-v3.2 --model-haiku minimax/minimax-m2.1
 
   # Use stdin for large prompts (e.g., git diffs, code review)
-  echo "Review this code..." | claudish --stdin --model x-ai/grok-code-fast-1
-  git diff | claudish --stdin --model openai/gpt-5-codex "Review these changes"
+  echo "Review this code..." | claudish --stdin --model g/gemini-2.0-flash
+  git diff | claudish --stdin --model openai/gpt-5.2 "Review these changes"
 
   # Monitor mode - understand how Claude Code works (requires real Anthropic API key)
   claudish --monitor --debug "analyze code structure"
@@ -61965,7 +61983,7 @@ var getRemoteProviders = () => [
     baseUrl: process.env.GEMINI_BASE_URL || "https://generativelanguage.googleapis.com",
     apiPath: "/v1beta/models/{model}:streamGenerateContent?alt=sse",
     apiKeyEnvVar: "GEMINI_API_KEY",
-    prefixes: ["g/", "gemini/", "google/"],
+    prefixes: ["g/", "gemini/"],
     capabilities: {
       supportsTools: true,
       supportsVision: true,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claudish",
-  "version": "3.1.1",
+  "version": "3.1.3",
   "description": "Run Claude Code with any model - OpenRouter, Ollama, LM Studio & local models",
   "type": "module",
   "main": "./dist/index.js",

package/skills/claudish-usage/SKILL.md

@@ -1,12 +1,12 @@
 ---
 name: claudish-usage
-description: CRITICAL - Guide for using Claudish CLI ONLY through sub-agents to run Claude Code with OpenRouter models (Grok, GPT-5, Gemini, MiniMax). NEVER run Claudish directly in main context unless user explicitly requests it. Use when user mentions external AI models, Claudish, OpenRouter, or alternative models. Includes mandatory sub-agent delegation patterns, agent selection guide, file-based instructions, and strict rules to prevent context window pollution.
+description: CRITICAL - Guide for using Claudish CLI ONLY through sub-agents to run Claude Code with any AI model (OpenRouter, Gemini, OpenAI, local models). NEVER run Claudish directly in main context unless user explicitly requests it. Use when user mentions external AI models, Claudish, OpenRouter, Gemini, OpenAI, Ollama, or alternative models. Includes mandatory sub-agent delegation patterns, agent selection guide, file-based instructions, and strict rules to prevent context window pollution.
 ---
 
 # Claudish Usage Skill
 
-**Version:** 1.1.0
-**Purpose:** Guide AI agents on how to use Claudish CLI to run Claude Code with OpenRouter models
+**Version:** 2.0.0
+**Purpose:** Guide AI agents on how to use Claudish CLI to run Claude Code with any AI model
 **Status:** Production Ready
 
 ## ⚠️ CRITICAL RULES - READ FIRST
@@ -151,50 +151,71 @@ Decision:
 
 ## Overview
 
-**Claudish** is a CLI tool that allows running Claude Code with any OpenRouter model (Grok, GPT-5, MiniMax, Gemini, etc.) by proxying requests through a local Anthropic API-compatible server.
+**Claudish** is a CLI tool that allows running Claude Code with any AI model via prefix-based routing. Supports OpenRouter (100+ models), direct Google Gemini API, direct OpenAI API, and local models (Ollama, LM Studio, vLLM, MLX).
 
 **Key Principle:** **ALWAYS** use Claudish through sub-agents with file-based instructions to avoid context window pollution.
 
 ## What is Claudish?
 
 Claudish (Claude-ish) is a proxy tool that:
-- ✅ Runs Claude Code with **any OpenRouter model** (not just Anthropic models)
+- ✅ Runs Claude Code with **any AI model** via prefix-based routing
+- ✅ Supports OpenRouter, Gemini, OpenAI, and local models
 - ✅ Uses local API-compatible proxy server
 - ✅ Supports 100% of Claude Code features
 - ✅ Provides cost tracking and model selection
 - ✅ Enables multi-model workflows
 
+## Model Routing
+
+| Prefix | Backend | Example |
+|--------|---------|---------|
+| _(none)_ | OpenRouter | `openai/gpt-5.2` |
+| `g/` `gemini/` | Google Gemini | `g/gemini-2.0-flash` |
+| `oai/` `openai/` | OpenAI | `oai/gpt-4o` |
+| `ollama/` | Ollama | `ollama/llama3.2` |
+| `lmstudio/` | LM Studio | `lmstudio/model` |
+| `http://...` | Custom | `http://localhost:8000/model` |
+
 **Use Cases:**
-- Run tasks with different AI models (Grok for speed, GPT-5 for reasoning, Gemini for vision)
+- Run tasks with different AI models (Grok for speed, GPT-5 for reasoning, Gemini for large context)
+- Use direct APIs for lower latency (Gemini, OpenAI)
+- Use local models for free, private inference (Ollama, LM Studio)
 - Compare model performance on same task
 - Reduce costs with cheaper models for simple tasks
-- Access models with specialized capabilities
 
 ## Requirements
 
 ### System Requirements
-- **OpenRouter API Key** - Required (set as `OPENROUTER_API_KEY` environment variable)
 - **Claudish CLI** - Install with: `npm install -g claudish` or `bun install -g claudish`
 - **Claude Code** - Must be installed
+- **At least one API key** (see below)
 
 ### Environment Variables
 
 ```bash
-# Required
-export OPENROUTER_API_KEY='sk-or-v1-...'  # Your OpenRouter API key
+# API Keys (at least one required)
+export OPENROUTER_API_KEY='sk-or-v1-...'  # OpenRouter (100+ models)
+export GEMINI_API_KEY='...'               # Direct Gemini API (g/ prefix)
+export OPENAI_API_KEY='sk-...'            # Direct OpenAI API (oai/ prefix)
+
+# Placeholder (required to prevent Claude Code dialog)
+export ANTHROPIC_API_KEY='sk-ant-api03-placeholder'
 
-# Optional (but recommended)
-export ANTHROPIC_API_KEY='sk-ant-api03-placeholder'  # Prevents Claude Code dialog
+# Custom endpoints (optional)
+export GEMINI_BASE_URL='https://...'      # Custom Gemini endpoint
+export OPENAI_BASE_URL='https://...'      # Custom OpenAI/Azure endpoint
+export OLLAMA_BASE_URL='http://...'       # Custom Ollama server
+export LMSTUDIO_BASE_URL='http://...'     # Custom LM Studio server
 
-# Optional - default model
-export CLAUDISH_MODEL='x-ai/grok-code-fast-1'  # or ANTHROPIC_MODEL
+# Default model (optional)
+export CLAUDISH_MODEL='openai/gpt-5.2'    # Default model
 ```
 
-**Get OpenRouter API Key:**
-1. Visit https://openrouter.ai/keys
-2. Sign up (free tier available)
-3. Create API key
-4. Set as environment variable
+**Get API Keys:**
+- OpenRouter: https://openrouter.ai/keys (free tier available)
+- Gemini: https://aistudio.google.com/apikey
+- OpenAI: https://platform.openai.com/api-keys
+- Local models: No API key needed
 
 ## Quick Start Guide
 
@@ -254,32 +275,25 @@ git diff | claudish --stdin --model openai/gpt-5-codex "Review these changes"
 
 ## Recommended Models
 
-**Top Models for Development (verified from OpenRouter):**
-
-1. **x-ai/grok-code-fast-1** - xAI's Grok (fast coding, visible reasoning)
-   - Category: coding
-   - Context: 256K
-   - Best for: Quick iterations, agentic coding
-
-2. **google/gemini-2.5-flash** - Google's Gemini (state-of-the-art reasoning)
-   - Category: reasoning
-   - Context: 1000K
-   - Best for: Complex analysis, multi-step reasoning
+**Top Models for Development (v3.1.1):**
 
-3. **minimax/minimax-m2** - MiniMax M2 (high performance)
-   - Category: coding
-   - Context: 128K
-   - Best for: General coding tasks
+| Model | Provider | Best For |
+|-------|----------|----------|
+| `openai/gpt-5.2` | OpenAI | **Default** - Most advanced reasoning |
+| `minimax/minimax-m2.1` | MiniMax | Budget-friendly, fast |
+| `z-ai/glm-4.7` | Z.AI | Balanced performance |
+| `google/gemini-3-pro-preview` | Google | 1M context window |
+| `moonshotai/kimi-k2-thinking` | MoonShot | Extended thinking |
+| `deepseek/deepseek-v3.2` | DeepSeek | Code specialist |
+| `qwen/qwen3-vl-235b-a22b-thinking` | Alibaba | Vision + reasoning |
 
-4. **openai/gpt-5** - OpenAI's GPT-5 (advanced reasoning)
-   - Category: reasoning
-   - Context: 128K
-   - Best for: Complex implementations, architecture decisions
+**Direct API Options (lower latency):**
 
-5. **qwen/qwen3-vl-235b-a22b-instruct** - Alibaba's Qwen (vision-language)
-   - Category: vision
-   - Context: 32K
-   - Best for: UI/visual tasks, design implementation
+| Model | Backend | Best For |
+|-------|---------|----------|
+| `g/gemini-2.0-flash` | Gemini | Fast tasks, large context |
+| `oai/gpt-4o` | OpenAI | General purpose |
+| `ollama/llama3.2` | Local | Free, private |
 
 **Get Latest Models:**
 ```bash
@@ -1294,5 +1308,5 @@ claudish --help-ai     # AI agent usage guide
 ---
 
 **Maintained by:** MadAppGang
-**Last Updated:** November 25, 2025
-**Skill Version:** 1.1.0
+**Last Updated:** January 5, 2026
+**Skill Version:** 2.0.0