assuremind 1.2.3 → 1.3.0

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 (11) hide show
  1. package/README.md +50 -1
  2. package/dist/cli/index.js +1146 -18485
  3. package/dist/index.js +56 -1912
  4. package/dist/index.mjs +56 -1807
  5. package/package.json +5 -5
  6. package/ui/dist/assets/{index-B2cmnLeA.js → index-Bkr40QZS.js} +1 -1
  7. package/ui/dist/favicon.svg +7 -7
  8. package/ui/dist/index.html +1 -1
  9. package/dist/cli/index.js.map +0 -1
  10. package/dist/index.js.map +0 -1
  11. package/dist/index.mjs.map +0 -1
package/README.md CHANGED
@@ -41,7 +41,17 @@ npx assuremind init # folders, config, Playwright browsers
41
41
  npx assuremind studio # opens http://localhost:4400
42
42
  ```
43
43
 
44
- > **Requirements:** Node.js >= 18 · macOS / Linux / Windows
44
+ ### System Requirements
45
+
46
+ | | Minimum | Recommended |
47
+ |---|---|---|
48
+ | **Node.js** | 18 LTS | 20 LTS or newer |
49
+ | **OS** | macOS / Linux / Windows 10+ | — |
50
+ | **RAM** | 4 GB | 8 GB+ |
51
+ | **Disk** | ~2 GB (app + Playwright browsers) | 5 GB+ |
52
+ | **Java** | — | JDK 17+ — only for Allure HTML reports ([adoptium.net](https://adoptium.net)) |
53
+
54
+ `npx assuremind init` installs the Playwright browsers. Using a **cloud AI provider** (Anthropic, OpenAI, Gemini, …) needs only an API key and the specs above. Running a **local model via Ollama** has heavier requirements — see below.
45
55
 
46
56
  ### Configure AI Provider
47
57
 
@@ -56,6 +66,45 @@ ANTHROPIC_MODEL=claude-sonnet.. OPENAI_MODEL=gpt-4o GOOGLE_MODEL=
56
66
 
57
67
  See `.env.example` for all 12 providers including Gemini, OpenAI, Anthropic, Ollama (local/free), etc.
58
68
 
69
+ ### Local & Free — Ollama
70
+
71
+ Run models entirely on your own machine (no API key, no cost). AssureMind uses the model to turn
72
+ plain-English steps into Playwright code, so a **code-tuned model with strong instruction-following**
73
+ works best.
74
+
75
+ ```bash
76
+ # 1. Install Ollama → https://ollama.com/download
77
+ ollama serve # starts the server on http://localhost:11434
78
+ ollama pull qwen2.5-coder:7b # 2. pull the recommended model
79
+ ```
80
+
81
+ ```bash
82
+ # 3. .env
83
+ AI_PROVIDER=ollama
84
+ OLLAMA_MODEL=qwen2.5-coder:7b
85
+ OLLAMA_BASE_URL=http://localhost:11434
86
+ AI_TIMEOUT=120 # local inference is slower than cloud APIs
87
+ ```
88
+
89
+ **Recommended models (by hardware):**
90
+
91
+ | Hardware | Model | Notes |
92
+ |----------|-------|-------|
93
+ | 16 GB+ RAM, GPU 8 GB+ VRAM | `qwen2.5-coder:14b` | Highest accuracy — best for complex steps / iframes |
94
+ | 16 GB RAM (GPU optional) ⭐ | `qwen2.5-coder:7b` | **Best balance** of quality & speed for this task |
95
+ | 8 GB RAM, no GPU | `llama3.2` (3B) / `qwen2.5-coder:3b` | Usable; good for simpler steps |
96
+ | ≤4–6 GB RAM | `llama3.2:1b` / `phi3:mini` | Lightweight; navigation & clicks only |
97
+
98
+ **Ollama system requirements:**
99
+
100
+ - **RAM is the floor** — a model needs roughly its file size **+ ≈2 GB free**. A 7B model (≈4.7 GB)
101
+ wants **8 GB free RAM minimum, 16 GB comfortable**; a 14B model wants **≈16 GB+**.
102
+ - **GPU is optional but transformative** — CPU-only works (hence `AI_TIMEOUT=120`), but a GPU with
103
+ **8 GB+ VRAM** (NVIDIA CUDA / Apple Silicon / modern AMD) makes 7B models fast. Apple Silicon (M-series) is excellent thanks to unified memory.
104
+ - **Disk** — a few GB per model; keep **10–20 GB free** if trying several.
105
+ - **Tips** — enable **MCP** (Settings → live page snapshots) for far more accurate locators with local
106
+ models. Avoid `deepseek-r1` here: its `<think>` reasoning output is slow and pollutes generated code.
107
+
59
108
  ---
60
109
 
61
110
  ## CLI