@austinthesing/magic-shell 0.2.20 → 0.2.22

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (5) hide show
  1. package/README.md +54 -27
  2. package/dist/cli.js +13810 -13707
  3. package/dist/index.js +4698 -1360
  4. package/dist/tui.js +13810 -13707
  5. package/package.json +13 -13
package/README.md CHANGED
@@ -7,7 +7,7 @@ Magic Shell is an open-source CLI tool that translates plain English (or any nat
7
7
  ## Features
8
8
 
9
9
  - **Natural Language Translation**: Describe what you want to do in plain English
10
- - **Multiple AI Providers**: OpenCode Zen (with free models!), OpenRouter, and custom models (LM Studio, Ollama, OpenAI-compatible)
10
+ - **Multiple AI Providers**: OpenCode Zen (with free models!), OpenRouter, Vercel AI Gateway, Cloudflare AI Gateway, Workers AI, and custom models (LM Studio, Ollama, OpenAI-compatible)
11
11
  - **Custom Model Support**: Add your own local or remote models with secure API key storage
12
12
  - **Project Context Aware**: Opt-in detection of package.json scripts, Makefile targets, etc.
13
13
  - **Interactive TUI Mode**: Full-featured terminal interface with themes
@@ -104,7 +104,7 @@ mshell
104
104
  | `msh --add-model` | Add custom model (LM Studio, Ollama, etc.) |
105
105
  | `msh --list-custom` | List custom models |
106
106
  | `msh --remove-model <id>` | Remove custom model |
107
- | `msh --provider <name>` | Set provider (opencode-zen or openrouter) |
107
+ | `msh --provider <name>` | Set provider (OpenCode Zen, OpenRouter, gateways, Workers AI, or custom) |
108
108
  | `msh --thinking <level>` | Set thinking level (off, low, medium, high) |
109
109
  | `msh --themes` | List available themes |
110
110
  | `msh --theme <name>` | Set color theme |
@@ -185,23 +185,23 @@ OpenCode Zen provides curated models optimized for coding tasks, including **fre
185
185
 
186
186
  **Free Models:**
187
187
 
188
- - `minimax-m2.5-free` - MiniMax's free model (default)
189
- - `ling-2.6-flash-free` - Ling's free flash model
190
- - `hy3-preview-free` - Hy3 preview model
191
- - `nemotron-3-super-free` - NVIDIA Nemotron free trial model
192
- - `trinity-large-preview-free` - Trinity large preview model
188
+ - `deepseek-v4-flash-free` - DeepSeek's free fast open-source model (default)
189
+ - `mimo-v2.5-free` - Xiaomi's free long-context MiMo model
190
+ - `north-mini-code-free` - North's free coding model
191
+ - `nemotron-3-ultra-free` - NVIDIA Nemotron free trial model
193
192
  - `big-pickle` - OpenCode stealth model
194
193
  - `gpt-5-nano` - OpenAI's free lightweight GPT model
195
194
 
196
195
  **Premium Models:**
197
196
 
198
- - Claude Sonnet 4.6, Claude Opus 4.7, Claude Haiku 4.5
199
- - Kimi K2.6, Kimi K2 Thinking
200
- - DeepSeek V4 Pro
201
- - GLM 5.1
197
+ - Claude Sonnet 4.6, Claude Opus 4.8, Claude Haiku 4.5, Claude Fable 5
198
+ - Kimi K2.6
199
+ - DeepSeek V4 Pro, DeepSeek V4 Flash
200
+ - GLM 5.2, GLM 5.1
201
+ - Qwen3.7 Max, Qwen3.7 Plus
202
202
  - MiMo V2.5, MiMo V2.5 Pro
203
203
  - MiniMax M2.7
204
- - Gemini 3.1 Pro, Gemini 3 Flash
204
+ - Gemini 3.5 Flash, Gemini 3.1 Pro, Gemini 3 Flash
205
205
  - GPT 5.5, GPT 5.5 Pro, GPT 5.4 Mini/Nano, GPT 5.3 Codex
206
206
  - And more...
207
207
 
@@ -211,17 +211,14 @@ Get your API key at: https://opencode.ai/auth
211
211
 
212
212
  Access to a wide variety of models from different providers.
213
213
 
214
- **Free Models:**
215
-
216
- - MiniMax M2.5 Free
217
-
218
- **Premium Models:**
214
+ **Open-source and open-weight highlights:**
219
215
 
220
- - MiMo V2.5, MiMo V2.5 Pro
221
- - Kimi K2.6, Kimi K2 Thinking
216
+ - Kimi K2.7 Code, Kimi K2.6
222
217
  - DeepSeek V4 Pro, DeepSeek V4 Flash
223
- - GLM 5.1, GLM 5 Turbo
224
- - MiniMax M2.7
218
+ - GLM 5.2, GLM 5.1, GLM 5 Turbo
219
+ - Qwen3.7 Max, Qwen3.7 Plus
220
+ - MiniMax M3, MiniMax M2.7
221
+ - MiMo V2.5, MiMo V2.5 Pro
225
222
  - And many more...
226
223
 
227
224
  Get your API key at: https://openrouter.ai/keys
@@ -303,7 +300,7 @@ Configuration is stored in `~/.magic-shell/config.json`.
303
300
  ```json
304
301
  {
305
302
  "provider": "opencode-zen",
306
- "defaultModel": "minimax-m2.5-free",
303
+ "defaultModel": "deepseek-v4-flash-free",
307
304
  "thinkingLevel": "low",
308
305
  "safetyLevel": "moderate",
309
306
  "dryRunByDefault": false,
@@ -326,11 +323,16 @@ Configuration is stored in `~/.magic-shell/config.json`.
326
323
 
327
324
  ### Environment Variables
328
325
 
329
- | Variable | Description |
330
- | ---------------------- | ----------------------------- |
331
- | `OPENCODE_ZEN_API_KEY` | API key for OpenCode Zen |
332
- | `OPENROUTER_API_KEY` | API key for OpenRouter |
333
- | `DEBUG_API=1` | Enable API response debugging |
326
+ | Variable | Description |
327
+ | -------------------------------- | ----------------------------------- |
328
+ | `OPENCODE_ZEN_API_KEY` | API key for OpenCode Zen |
329
+ | `OPENROUTER_API_KEY` | API key for OpenRouter |
330
+ | `AI_GATEWAY_API_KEY` | API key for Vercel AI Gateway |
331
+ | `CLOUDFLARE_AI_GATEWAY_API_KEY` | API key/token for Cloudflare Gateway |
332
+ | `CLOUDFLARE_API_TOKEN` | API token for Workers AI |
333
+ | `CLOUDFLARE_ACCOUNT_ID` | Account ID for Cloudflare providers |
334
+ | `CLOUDFLARE_AI_GATEWAY_ID` | Gateway ID for Cloudflare Gateway |
335
+ | `DEBUG_API=1` | Enable API response debugging |
334
336
 
335
337
  ## Themes
336
338
 
@@ -583,6 +585,31 @@ MIT License - see [LICENSE](LICENSE) for details.
583
585
 
584
586
  ## Release Notes
585
587
 
588
+ ### v0.2.22 - Model Catalog and Release Prep
589
+
590
+ - Refreshed OpenCode Zen and OpenRouter model registries with current June 2026 open-source/open-weight entries
591
+ - Updated the default free Zen model to `deepseek-v4-flash-free`
592
+ - Added Zen routing support for Qwen models on the Messages endpoint
593
+ - Updated release docs and website provider surfaces for the refreshed provider/model set
594
+
595
+ ### v0.2.20 - Thinking Controls and Zen Free Models
596
+ - Added `--thinking <level>` and `thinkingLevel` config support (`off`, `low`, `medium`, `high`)
597
+ - Added provider-specific reasoning controls for supported OpenRouter, OpenCode Zen, gateway, and custom models
598
+ - Updated OpenCode Zen free models and default model references to `minimax-m2.5-free`
599
+ - Disabled custom thinking parameters when the selected model/provider does not support them
600
+
601
+ ### v0.2.19 - AI Gateway Providers
602
+ - Added Vercel AI Gateway, Cloudflare AI Gateway, and Cloudflare Workers AI providers
603
+ - Added gateway provider setup, switching, model listing, API-key handling, and environment variables
604
+ - Refreshed OpenCode Zen and OpenRouter registries with current OpenAI, Anthropic, Kimi, DeepSeek, GLM, MiniMax, and MiMo model IDs
605
+ - Added provider-scoped model helpers so model selection follows the active provider
606
+
607
+ ### v0.2.18 - Model Registry Refresh
608
+ - Added latest MiMo V2.5 and MiMo V2.5 Pro entries for OpenCode Zen
609
+ - Added latest Xiaomi MiMo V2.5 and MiMo V2.5 Pro entries for OpenRouter
610
+ - Removed the older OpenRouter MiMo V2 Flash free entry from docs and model listings
611
+ - Updated runtime dependencies for the published CLI package
612
+
586
613
  ### v0.2.17 - Dependency Maintenance
587
614
  - Updated published CLI dependencies to current releases
588
615
  - Fixed TUI input typing compatibility after the `@opentui/core` upgrade