llms-py 3.0.0b9__tar.gz → 3.0.1__tar.gz

{llms_py-3.0.0b9/llms_py.egg-info → llms_py-3.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llms-py
-Version: 3.0.0b9
+Version: 3.0.1
 Summary: A lightweight CLI tool and OpenAI-compatible server for querying multiple Large Language Model (LLM) providers
 Home-page: https://github.com/ServiceStack/llms
 Author: ServiceStack

llms_py-3.0.1/llms/extensions/app/README.md

@@ -0,0 +1,20 @@
+# App Extension
+
+This extension provides the core application logic and data persistence for the LLMs platform.
+
+## Data Storage & Architecture
+
+### Server-Side SQLite Migration
+The application has migrated from client-side IndexedDB storage to a robust server-side SQLite solution. This architectural shift ensures better data consistency, improved performance, and enables multi-device access to your chat history.
+
+### Asset Management
+To keep the database efficient and portable, binary assets (images, audio, etc.) are not stored directly in the SQLite database. Instead:
+- All generated assets are stored in the local file system cache at `~/.llms/cache`.
+- The database stores only **relative URLs** pointing to these assets.
+- This approach allows for efficient caching and serving of static media.
+
+### Concurrency Model
+To ensure data integrity and high performance without complex locking mechanisms, the system utilizes a **single background thread** for managing all write operations to the database. This design improves concurrency handling and eliminates database locking issues during high-load scenarios.
+
+### Multi-Tenancy & Security
+When authentication is enabled, data isolation is automatically enforced. All core tables, including `threads` and `requests`, are scoped to the authenticated user, ensuring that users can only access their own data.

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/app/__init__.py

@@ -25,16 +25,6 @@ def install(ctx):
                 g_db = AppDB(ctx, db_path)
                 ctx.register_shutdown_handler(g_db.close)
 
-                threads_json = "/home/mythz/src/ServiceStack/llms/llms/extensions/app/threads.json"
-                requests_json = "/home/mythz/src/ServiceStack/llms/llms/extensions/app/requests.json"
-                with open(threads_json) as f:
-                    threads = json.load(f)["threads"]
-                    threads.reverse()
-                with open(requests_json) as f:
-                    requests = json.load(f)["requests"]
-                    requests.reverse()
-                # g_db.import_db(threads, requests)
-
             except Exception as e:
                 ctx.err("Failed to init AppDB", e)
         return g_db
@@ -60,7 +50,7 @@ def install(ctx):
         return to
 
     def thread_dto(row):
-        return row and to_dto(row, ["messages", "modalities", "args", "modelInfo", "stats"])
+        return row and to_dto(row, ["messages", "modalities", "args", "modelInfo", "stats", "metadata"])
 
     def request_dto(row):
         return row and to_dto(row, ["usage"])
@@ -174,12 +164,13 @@ def install(ctx):
         if not thread:
             raise Exception("Thread not found")
 
+        metadata = thread.get("metadata", {})
         chat = {
             "model": thread.get("model"),
             "messages": thread.get("messages"),
             "modalities": thread.get("modalities"),
             "systemPrompt": thread.get("systemPrompt"),
-            "metadata": thread.get("metadata", {}),
+            "metadata": metadata,
         }
         for k, v in thread.get("args", {}).items():
             if k in ctx.request_args:
@@ -189,7 +180,8 @@ def install(ctx):
             "chat": chat,
             "user": user,
             "threadId": id,
-            "tools": chat.get("metadata").get("tools", "all"),
+            "metadata": metadata,
+            "tools": metadata.get("tools", "all"),
         }
 
         # execute chat in background thread
@@ -341,13 +333,22 @@ def install(ctx):
 
     ctx.register_chat_request_filter(chat_request)
 
-    async def tool_request(tool_chat, context):
-        ctx.dbg("tool_request")
+    async def tool_request(chat_request, context):
+        messages = chat_request.get("messages", [])
+        ctx.dbg(f"tool_request: messages {len(messages)}")
         thread_id = context.get("threadId", None)
         if not thread_id:
             ctx.dbg("Missing threadId")
             return
         user = context.get("user", None)
+        await g_db.update_thread_async(
+            thread_id,
+            {
+                "messages": messages,
+            },
+            user=user,
+        )
+
         completed_at = g_db.get_thread_column(thread_id, "completedAt", user=user)
         if completed_at:
             context["completed"] = True

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/app/db.py

@@ -91,6 +91,7 @@ class AppDB:
                 "publishedAt": "TIMESTAMP",
                 "startedAt": "TIMESTAMP",
                 "completedAt": "TIMESTAMP",
+                "metadata": "JSON",
                 "error": "TEXT",
                 "ref": "TEXT",
             },
@@ -444,7 +445,7 @@ class AppDB:
         event.wait()
         return ret[0]
 
-    def prepare_thread(self, thread, id=None):
+    def prepare_thread(self, thread, id=None, user=None):
         now = datetime.now()
         if id:
             thread["id"] = id
@@ -454,19 +455,19 @@ class AppDB:
         if "messages" in thread:
             for m in thread["messages"]:
                 self.ctx.cache_message_inline_data(m)
-        return thread
+        return with_user(thread, user=user)
 
     def create_thread(self, thread: Dict[str, Any], user=None):
-        return self.insert("thread", with_user(self.prepare_thread(thread), user=user))
+        return self.insert("thread", self.prepare_thread(thread, user=user))
 
     async def create_thread_async(self, thread: Dict[str, Any], user=None):
-        return await self.insert_async("thread", with_user(self.prepare_thread(thread), user=user))
+        return await self.insert_async("thread", self.prepare_thread(thread, user=user))
 
     def update_thread(self, id, thread: Dict[str, Any], user=None):
-        return self.update("thread", with_user(self.prepare_thread(thread, id), user=user))
+        return self.update("thread", self.prepare_thread(thread, id, user=user))
 
     async def update_thread_async(self, id, thread: Dict[str, Any], user=None):
-        return await self.update_async("thread", with_user(self.prepare_thread(thread, id), user=user))
+        return await self.update_async("thread", self.prepare_thread(thread, id, user=user))
 
     def delete_thread(self, id, user=None, callback=None):
         sql_where, params = self.get_user_filter(user, {"id": id})

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/app/ui/index.mjs

@@ -219,7 +219,7 @@ const ThreadsSidebar = {
         }
 
         const createNewThread = async () => {
-            ctx.threads.startNewThread({ title: 'New Chat', model: ctx.chat.getSelectedModel() })
+            ctx.threads.startNewThread({ title: 'New Chat', model: ctx.chat.getSelectedModel(), redirect: true })
         }
 
         const goToInitialState = () => {

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/app/ui/threadStore.mjs

@@ -1,5 +1,5 @@
 import { ref, computed } from 'vue'
-import { appendQueryString, createErrorStatus } from '@servicestack/client'
+import { appendQueryString } from '@servicestack/client'
 
 /**
  * Returns an ever-increasing unique integer id.
@@ -21,11 +21,6 @@ const isLoading = ref(false)
 let ctx = null
 let ext = null
 
-// Generate unique thread ID
-function generateThreadId() {
-    return Date.now().toString()
-}
-
 function setError(error, msg = null) {
     ctx?.setError(error, msg)
 }
@@ -170,7 +165,7 @@ async function redoMessageFromThread(threadId, timestamp) {
     // Find the index of the message to redo
     const messageIndex = thread.messages.findIndex(m => m.timestamp === timestamp)
     if (messageIndex === -1) {
-        setError(createErrorStatus(`Message not found for timestamp ${timestamp}`))
+        setError({ message: `Message not found for timestamp ${timestamp}` })
         return
     }
 
@@ -188,7 +183,8 @@ async function redoMessageFromThread(threadId, timestamp) {
     const updatedMessages = thread.messages.slice(0, messageIndex + 1)
 
     // Update the thread with the new messages
-    const api = await queueChat(threadId, { messages: updatedMessages })
+    const request = { messages: updatedMessages }
+    const api = await queueChat({ request, thread })
     if (api.response) {
         replaceThread(api.response)
     } else {
@@ -311,7 +307,7 @@ function getLatestCachedThread() {
     return threads.value[0]
 }
 
-async function startNewThread({ title, model }) {
+async function startNewThread({ title, model, redirect }) {
     if (!model) {
         console.error('No model selected')
         return
@@ -337,8 +333,10 @@ async function startNewThread({ title, model }) {
     })
 
     console.log('newThread', newThread, model)
-    // Navigate to the new thread URL
-    ctx.to(`/c/${newThread.id}`)
+    if (redirect) {
+        // Navigate to the new thread URL
+        ctx.to(`/c/${newThread.id}`)
+    }
 
     // Get the thread to check for duplicates
     let thread = await getThread(newThread.id)
@@ -346,12 +344,19 @@ async function startNewThread({ title, model }) {
     return thread
 }
 
-async function queueChat(threadId, body, options = {}) {
-    const api = await ctx.postJson(`/ext/app/threads/${threadId}/chat`, {
+async function queueChat(ctxRequest, options = {}) {
+    if (!ctxRequest.request) return ctx.createErrorResult({ message: 'No request provided' })
+    if (!ctxRequest.thread) return ctx.createErrorResult({ message: 'No thread provided' })
+    if (!ctxRequest.request.metadata) {
+        ctxRequest.request.metadata = {}
+    }
+    ctx.chatRequestFilters.forEach(f => f(ctxRequest))
+    const { thread, request } = ctxRequest
+    const api = await ctx.postJson(`/ext/app/threads/${thread.id}/chat`, {
         ...options,
-        body: typeof body == 'string'
-            ? body
-            : JSON.stringify(body),
+        body: typeof request == 'string'
+            ? request
+            : JSON.stringify(request),
     })
     return api
 }
@@ -380,7 +385,6 @@ export function useThreadStore() {
         clearCurrentThread,
         getGroupedThreads,
         getLatestCachedThread,
-        generateThreadId,
         startNewThread,
         replaceThread,
         queueChat,

llms_py-3.0.1/llms/extensions/core_tools/CALCULATOR.md

@@ -0,0 +1,32 @@
+# Calculator
+
+A powerful and safe mathematical expression evaluator with a rich web interface.
+
+## Features
+
+### 🖥️ UX Friendly Interface
+Experience a clean, modern interface designed for efficiency. The UI is fully responsive and supports dark mode, seamlessly integrating with the rest of the application.
+
+### 💾 Persistent History
+Never lose track of your calculations. The Calculator automatically saves your history to `localStorage`, ensuring your previous expressions and results are preserved between sessions.
+
+### ⚡ 1-Click Interaction
+Streamline your workflow with interactive history items:
+- **Load & Copy**: Click on any past expression or answer to instantly load it into the input field and copy it to your clipboard.
+- **Visual Feedback**: Temporary checkmarks confirm successful copy actions.
+
+### ⌨️ Keyboard-Free Access
+While full keyboard support is available, you can perform complex calculations entirely via the UI:
+- **Numbers & Constants**: Quick access to digits and mathematical constants like `pi`, `e`, `inf`.
+- **Operators**: Comprehensive set of buttons for arithmetic (`+`, `-`, `*`, `/`, `%`, `^`) and boolean logic (`and`, `or`, `not`).
+- **Functions**: One-click insertion or wrapping of selection for all supported math functions.
+
+### 🐍 Python Math Support
+Unlock the power of Python's math library directly in the browser.
+- **Math Functions**: Support for `sin`, `cos`, `tan`, `sqrt`, `log`, `factorial`, and many more.
+- **Statistics**: Built-in functions for `mean`, `median`, `stdev`, and `variance`.
+
+### 🛡️ Safe Evaluation
+Security is a priority. Instead of using Python's unsafe `eval()`, the Calculator uses a robust **AST (Abstract Syntax Tree) evaluator**.
+- **Restricted Environment**: Only allowed mathematical operations and functions are executed.
+- **No Side Effects**: Prevents arbitrary code execution, making it safe to evaluate expressions from untrusted sources.

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/core_tools/__init__.py

@@ -523,7 +523,7 @@ def install(ctx):
     # Examples of registering tools using automatic definition generation
     ctx.register_tool(memory_read)
     ctx.register_tool(memory_write)
-    ctx.register_tool(semantic_search)
+    # ctx.register_tool(semantic_search) # TODO: implement
     ctx.register_tool(read_file)
     ctx.register_tool(write_file)
     ctx.register_tool(list_directory)

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/core_tools/ui/index.mjs

@@ -381,7 +381,7 @@ const CalcPage = {
                                     type="button"
                                     @click="insert(num)"
                                     class="px-3 py-1 bg-gray-100 dark:bg-gray-800 hover:bg-blue-100 dark:hover:bg-blue-900/30 text-gray-700 dark:text-gray-300 hover:text-blue-700 dark:hover:text-blue-300 border border-gray-200 dark:border-gray-700 rounded text-sm font-mono transition-colors"
-                                    title="Click to insert"
+                                    :title="'insert number ' + num"
                                 >
                                     {{ num }}
                                 </button>
@@ -392,7 +392,7 @@ const CalcPage = {
                                     type="button"
                                     @click="insert(c)"
                                     class="px-3 py-1 bg-gray-100 dark:bg-gray-800 hover:bg-blue-100 dark:hover:bg-blue-900/30 text-gray-700 dark:text-gray-300 hover:text-blue-700 dark:hover:text-blue-300 border border-gray-200 dark:border-gray-700 rounded text-sm font-mono transition-colors"
-                                    title="Click to insert"
+                                    :title="'insert constant ' + c"
                                 >
                                     {{ c }}
                                 </button>
@@ -408,7 +408,7 @@ const CalcPage = {
                                     type="button"
                                     @click="insert(op)"
                                     class="px-3 py-1 bg-gray-100 dark:bg-gray-800 hover:bg-blue-100 dark:hover:bg-blue-900/30 text-gray-700 dark:text-gray-300 hover:text-blue-700 dark:hover:text-blue-300 border border-gray-200 dark:border-gray-700 rounded text-sm font-mono transition-colors"
-                                    title="Click to insert"
+                                    :title="'insert operator ' + op"
                                 >
                                     {{ op }}
                                 </button>
@@ -425,7 +425,7 @@ const CalcPage = {
                                     type="button"
                                     @click="wrapWithFunction(func)"
                                     class="px-3 py-1 bg-gray-100 dark:bg-gray-800 hover:bg-purple-100 dark:hover:bg-purple-900/30 text-gray-700 dark:text-gray-300 hover:text-purple-700 dark:hover:text-purple-300 border border-gray-200 dark:border-gray-700 rounded text-sm font-mono transition-colors"
-                                    title="Click to insert"
+                                    :title="'use function ' + func"
                                 >
                                     {{ func }}
                                 </button>

llms_py-3.0.1/llms/extensions/gallery/README.md

@@ -0,0 +1,61 @@
+# Gallery Extension
+
+The Gallery extension intercepts all generated image, audio & file assets and uploaded files in `~/.llms/cache` file storage whose metadata is maintained in a SQLite database.
+
+Dedicated UIs are available for quickly browsing and navigating or generated images / audio files with optimized UIs for viewing portrait, square and landscape images.
+
+## Generated Asset Interception
+
+The Gallery extension automatically monitors the creation of new cache entries. Whenever a file is saved to the LLMs cache (located at `~/.llms/cache`), the extension captures its metadata and stores it in the gallery database.
+
+This includes:
+- **Generated Images**: Images created by AI models (e.g., DALL-E, Stable Diffusion).
+- **Generated Audio**: Audio files generated by text-to-speech or audio models.
+- **Uploaded Files**: Any files uploaded through the UI.
+
+All metadata is stored in a dedicated SQLite database located at `~/.llms/user/default/gallery/gallery.sqlite`, in the `media` table.
+
+## User Interface
+
+The Gallery UI provides a rich, interactive way to explore your generated assets. You can access it via the **Gallery** tab in the sidebar or by navigating to `/gallery`.
+
+### Image Gallery
+
+The image view offers a responsive grid layout optimized for different aspect ratios.
+
+- **Filtering**:
+    - **By Format**: Easily switch between **Portrait**, **Square**, and **Landscape** views to see images in their best light.
+    - **Search**: Real-time search by prompt, model name, or other metadata.
+- **Interactions**:
+    - **Lightbox**: Click any image to view it in full screen.
+    - **Details**: View comprehensive metadata including the prompt used, generation model, dimensions, file size, creation date, and generation cost.
+    - **Download**: extensive download options.
+    - **Remix**: Quickly re-use the prompt and settings of an existing image to generate a new one.
+    - **Delete**: Remove unwanted images from the gallery.
+
+### Audio Gallery
+
+The audio view presents a list layout designed for easy listening and management.
+
+- **Playback**: Integrated audio player to preview generated sounds directly in the list.
+- **Metadata**: Displays the caption/prompt, model, and creation time.
+- **Actions**:
+    - **Remix**: Regenerate audio using the same prompt.
+    - **Delete**: Remove audio files.
+
+## Storage Data model
+
+The `media` table tracks extensive information about each asset to support the search and filtering capabilities:
+
+| Column | Description |
+|---|---|
+| `url` | Relative path to the file in `~/.llms/cache` |
+| `type` | Asset type (image, audio, video) |
+| `prompt` | The prompt used to generate the asset |
+| `model` | The AI model used |
+| `aspect_ratio` | Aspect ratio (e.g., "1:1", "16:9") |
+| `cost` | Generation cost |
+| `metadata` | Additional JSON metadata |
+| `created` | Timestamp of creation |
+
+This local database ensures your gallery remains fast and responsive, even with a large collection of generated assets.

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/gallery/ui/index.mjs

@@ -390,6 +390,7 @@ const GalleryPage = {
             ctx.threads.startNewThread({
                 title: selected.prompt,
                 model: ctx.chat.getSelectedModel(),
+                redirect: true,
             })
         }
 

llms_py-3.0.1/llms/extensions/katex/README.md

@@ -0,0 +1,39 @@
+# KaTeX Extension
+
+This extension enables beautiful rendering of LaTeX math expressions in AI responses using [KaTeX](https://katex.org/). It integrates automatically with the markdown parser to render math equations in both inline and block formats.
+
+## Features
+
+- **Fast Rendering**: Uses KaTeX for high-performance rendering of math expressions.
+- **Inline Math**: Renders math within text using `$` or `$$` delimiters.
+- **Block Math**: Renders complex equations in their own block using `$` or `$$` delimiters across multiple lines.
+- **Auto-Integration**: Automatically extends the `marked` parser used in the application.
+
+## Usage
+
+The extension supports standard LaTeX math syntax.
+
+### Inline Math
+
+Surround your LaTeX expression with single `$` (for inline style) or double `$$` (for display style) delimiters.
+
+**Example:**
+`The quadratic formula is $x = \frac{-b \pm \sqrt{b^2 - 4ac}}{2a}$.`
+
+### Block Math
+
+For larger equations or when you want the math to be displayed on its own line, use block syntax by placing the delimiters on separate lines. Standard usage is to use double `$$` delimiters.
+
+**Example:**
+```latex
+$$
+\int_{-\infty}^{\infty} e^{-x^2} dx = \sqrt{\pi}
+$$
+```
+
+## Configuration
+
+The extension automatically registers:
+- **Import Maps**: Loads `katex.min.mjs` for the frontend.
+- **CSS**: Injects `katex.min.css` for styling.
+- **Markdown Extension**: Adds a custom tokenizer and renderer to `marked` to detect and render LaTeX patterns.

llms_py-3.0.1/llms/extensions/system_prompts/README.md

@@ -0,0 +1,22 @@
+# System Prompts Extension
+
+This extension configures AI requests with a library of **over 200+** awesome curated system prompts that can be selected from the UI.
+
+## Custom System Prompts
+
+You can also maintain your own library of system prompts which can be maintained for all anonymous users at:
+`~/.llms/user/default/system-prompts.json`
+
+Or for signed in users at:
+`~/.llms/user/<github-user>/system-prompts.json`
+
+The JSON file should contain an array of Prompt objects, e.g:
+
+```json
+[
+    {
+        "name": "Helpful Assistant",
+        "prompt": "You are a helpful assistant."
+    }
+]
+```

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/system_prompts/ui/index.mjs

@@ -142,7 +142,7 @@ const SystemPromptEditor = {
                     System Prompt
                 </label>
                 <div v-if="hasMessages" class="text-sm text-gray-500 dark:text-gray-400">
-                    {{ !threadSystemPrompt ? '' : prompts.find(x => x.value === threadSystemPrompt)?.name || 'Custom' }}
+                    {{ !ext.prefs.systemPrompt ? '' : prompts.find(x => x.value === ext.prefs.systemPrompt)?.name || 'Custom' }}
                 </div>
                 <div v-else class="mb-2 relative" ref="containerRef">
                     <div class="flex items-center gap-2">
@@ -154,16 +154,16 @@ const SystemPromptEditor = {
                             <svg class="size-4 text-gray-500 dark:text-gray-400" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path fill="currentColor" d="M19 6.41L17.59 5L12 10.59L6.41 5L5 6.41L10.59 12L5 17.59L6.41 19L12 13.41L17.59 19L19 17.59L13.41 12z"/></svg>
                         </button>
                         <button type="button" 
-                            @click="showFinder = !showFinder"
+                            @click="ext.setPrefs({ showFinder: !ext.prefs.showFinder })"
                             class="inline-flex items-center gap-x-1.5 rounded-md bg-white dark:bg-gray-900 px-2.5 py-1.5 text-sm font-medium text-gray-700 dark:text-gray-300 shadow-sm border border-gray-300 dark:border-gray-600 hover:bg-gray-50 dark:hover:bg-gray-800">
                             Explore Prompts
                         </button>
                     </div>
-                    <PromptFinder v-model="showFinder" :prompts="prompts" @select="onSelect" />
+                    <PromptFinder v-model="ext.prefs.showFinder" :prompts="prompts" @select="onSelect" />
                 </div>
             </div>
             <div v-if="hasMessages" class="w-full rounded-md border border-gray-300 dark:border-gray-600 bg-white dark:bg-gray-900 text-gray-900 dark:text-gray-100 px-3 py-2 text-sm">
-                {{threadSystemPrompt || 'No System Prompt was used' }}
+                {{$threads.currentThread.value?.systemPrompt || 'No System Prompt was used' }}
             </div>
             <div v-else>
                 <textarea
@@ -186,32 +186,29 @@ const SystemPromptEditor = {
         /**@type {AppContext} */
         const ctx = inject('ctx')
         const containerRef = ref()
-        const showFinder = ref(false)
-        const prefs = ext.getPrefs()
         const hasMessages = computed(() => ctx.threads.currentThread.value?.messages?.length > 0)
-        const threadSystemPrompt = computed(() => ctx.threads.currentThread.value?.systemPrompt || '')
         const selected = computed(() =>
             props.prompts.find(x => x.value === props.modelValue) ?? { name: "Custom", value: props.modelValue })
 
         function onSelect(prompt) {
+            ext.setPrefs({ prompt: prompt }) // {"id","name","value"}
             emit('update:modelValue', prompt.value)
         }
 
         function closeFinder(e) {
-            if (showFinder.value && containerRef.value && !containerRef.value.contains(e.target)) {
-                showFinder.value = false
+            if (ext.prefs.showFinder && containerRef.value && !containerRef.value.contains(e.target)) {
+                ext.setPrefs({ showFinder: false })
             }
         }
 
-        watch(() => props.modelValue, promptValue => {
-            prefs.prompt = selected.value
-            ext.setPrefs(prefs)
+        watch(() => props.modelValue, systemPrompt => {
+            ext.setPrefs({ systemPrompt })
         })
 
         onMounted(() => {
             document.addEventListener('click', closeFinder)
-            if (prefs.prompt) {
-                emit('update:modelValue', prefs.prompt.value)
+            if (ext.prefs.prompt) {
+                emit('update:modelValue', ext.prefs.prompt)
             }
         })
         onUnmounted(() => {
@@ -219,11 +216,10 @@ const SystemPromptEditor = {
         })
 
         return {
-            threadSystemPrompt,
+            ext,
             hasMessages,
             selected,
             containerRef,
-            showFinder,
             onSelect,
         }
     }
@@ -238,9 +234,13 @@ export default {
             PromptFinder,
             SystemPromptEditor,
             SystemPromptsPanel: {
-                template: `<SystemPromptEditor :prompts="$state.prompts" v-model="$state.selectedPrompt" />`,
+                template: `<SystemPromptEditor :prompts="ext.state.prompts" v-model="ext.prefs.prompt" />`,
+                setup() {
+                    return { ext }
+                }
             }
         })
+        ext.setPrefs({ systemPrompt: '' })
 
         ctx.setTopIcons({
             system_prompts: {
@@ -251,12 +251,6 @@ export default {
             }
         })
 
-        ctx.createThreadFilters.push(thread => {
-            const prefs = ext.getPrefs()
-            thread.systemPrompt = prefs?.prompt?.value || ""
-            console.log('createThreadFilters', prefs, thread)
-        })
-
         ctx.chatRequestFilters.push(({ request, thread }) => {
 
             const hasSystemPrompt = request.messages.find(x => x.role === 'system')
@@ -265,11 +259,12 @@ export default {
                 return
             }
 
-            if (thread.systemPrompt) {
+            // Only add the selected system prompt for new requests
+            if (ext.prefs.systemPrompt && request.messages.length <= 1) {
                 // add message to start
                 request.messages.unshift({
                     role: 'system',
-                    content: thread.systemPrompt
+                    content: ext.prefs.systemPrompt
                 })
             }
         })
@@ -280,6 +275,6 @@ export default {
     async load(ctx) {
         const api = await ext.getJson(`/prompts.json`)
         const prompts = api.response || []
-        ctx.setState({ prompts })
+        ext.setState({ prompts })
     }
 }

{llms_py-3.0.0b9 → llms_py-3.0.1}/llms/extensions/system_prompts/ui/prompts.json

@@ -645,7 +645,7 @@
         "value": "I want you to act as an biblical translator. I will speak to you in english and you will translate it and answer in the corrected and improved version of my text, in a biblical dialect. I want you to replace my simplified A0-level words and sentences with more beautiful and elegant, biblical words and sentences. Keep the meaning same. I want you to only reply the correction, the improvements and nothing else, do not write explanations. My first sentence is \"Hello, World!\""
     },
     {
-        "id": "chess-player",
+        "id": "chess-player2",
         "name": "Act as an Chess Player",
         "value": "I want you to act as a rival chess player. I We will say our moves in reciprocal order. In the beginning I will be white. Also please don't explain your moves to me because we are rivals. After my first message i will just write my move. Don't forget to update the state of the board in your mind as we make moves. My first move is e4."
     },
@@ -740,7 +740,7 @@
         "value": "I want you to act as a Graphviz DOT generator, an expert to create meaningful diagrams. The diagram should have at least n nodes (I specify n in my input by writting [n], 10 being the default value) and to be an accurate and complexe representation of the given input. Each node is indexed by a number to reduce the size of the output, should not include any styling, and with layout=neato, overlap=false, node [shape=rectangle] as parameters. The code should be valid, bugless and returned on a single line, without any explanation. Provide a clear and organized diagram, the relationships between the nodes have to make sense for an expert of that input. My first diagram is: \"The water cycle [8]\"."
     },
     {
-        "id": "life-coach",
+        "id": "life-coach2",
         "name": "Act as a Life Coach",
         "value": "I want you to act as a Life Coach. Please summarize this non-fiction book, [title] by [author]. Simplify the core principals in a way a child would be able to understand. Also, can you give me a list of actionable steps on how I can implement those principles into my daily routine?"
     },
@@ -840,7 +840,7 @@
         "value": "I want you to act as a Japanese Kanji quiz machine. Each time I ask you for the next question, you are to provide one random Japanese kanji from JLPT N5 kanji list and ask for its meaning. You will generate four options, one correct, three wrong. The options will be labeled from A to D. I will reply to you with one letter, corresponding to one of these labels. You will evaluate my each answer based on your last question and tell me if I chose the right option. If I chose the right label, you will congratulate me. Otherwise you will tell me the right answer. Then you will ask me the next question."
     },
     {
-        "id": "note-taking-assistant",
+        "id": "note-taking-assistant2",
         "name": "Act as a note-taking assistant",
         "value": "I want you to act as a note-taking assistant for a lecture. Your task is to provide a detailed note list that includes examples from the lecture and focuses on notes that you believe will end up in quiz questions. Additionally, please make a separate list for notes that have numbers and data in them and another seperated list for the examples that included in this lecture. The notes should be concise and easy to read."
     },
@@ -975,7 +975,7 @@
         "value": "You are the \"Architect Guide,\" specialized in assisting programmers who are experienced in individual module development but are looking to enhance their skills in understanding and managing entire project architectures. Your primary roles and methods of guidance include:\n\n- **Basics of Project Architecture**: Start with foundational knowledge, focusing on principles and practices of inter-module communication and standardization in modular coding. - **Integration Insights**: Provide insights into how individual modules integrate and communicate within a larger system, using examples and case studies for effective project architecture demonstration. - **Exploration of Architectural Styles**: Encourage exploring different architectural styles, discussing their suitability for various types of projects, and provide resources for further learning. - **Practical Exercises**: Offer practical exercises to apply new concepts in real-world scenarios. - **Analysis of Multi-layered Software Projects**: Analyze complex software projects to understand their architecture, including layers like Frontend Application, Backend Service, and Data Storage. - **Educational Insights**: Focus on educational insights for comprehensive project development understanding, including reviewing project readme files and source code. - **Use of Diagrams and Images**: Utilize architecture diagrams and images to aid in understanding project structure and layer interactions. - **Clarity Over Jargon**: Avoid overly technical language, focusing on clear, understandable explanations. - **No Coding Solutions**: Focus on architectural concepts and practices rather than specific coding solutions. - **Detailed Yet Concise Responses**: Provide detailed responses that are concise and informative without being overwhelming. - **Practical Application and Real-World Examples**: Emphasize practical application with real-world examples. - **Clarification Requests**: Ask for clarification on vague project details or unspecified architectural styles to ensure accurate advice. - **Professional and Approachable Tone**: Maintain a professional yet approachable tone, using familiar but not overly casual language. - **Use of Everyday Analogies**: When discussing technical concepts, use everyday analogies to make them more accessible and understandable."
     },
     {
-        "id": "chatgpt-prompt-generator",
+        "id": "chatgpt-prompt-generator2",
         "name": "Act as ChatGPT Prompt Generator",
         "value": "Let's refine the process of creating high-quality prompts together. Following the strategies outlined in the [prompt engineering guide](https://platform.openai.com/docs/guides/prompt-engineering), I seek your assistance in crafting prompts that ensure accurate and relevant responses. Here's how we can proceed:\n\n1. **Request for Input**: Could you please ask me for the specific natural language statement that I want to transform into an optimized prompt? 2. **Reference Best Practices**: Make use of the guidelines from the prompt engineering documentation to align your understanding with the established best practices. 3. **Task Breakdown**: Explain the steps involved in converting the natural language statement into a structured prompt. 4. **Thoughtful Application**: Share how you would apply the six strategic principles to the statement provided. 5. **Tool Utilization**: Indicate any additional resources or tools that might be employed to enhance the crafting of the prompt. 6. **Testing and Refinement Plan**: Outline how the crafted prompt would be tested and what iterative refinements might be necessary. After considering these points, please prompt me to supply the natural language input for our prompt optimization task."
     },
@@ -1060,7 +1060,7 @@
         "value": "Using WebPilot, create an outline for an article that will be 2,000 words on the keyword “Best SEO Prompts” based on the top 10 results from Google. Include every relevant heading possible. Keep the keyword density of the headings high.<br/> For each section of the outline, include the word count.<br/> Include FAQs section in the outline too, based on people also ask section from Google for the keyword.<br/> This outline must be very detailed and comprehensive, so that I can create a 2,000 word article from it. Generate a long list of LSI and NLP keywords related to my keyword. Also include any other words related to the keyword.<br/> Give me a list of 3 relevant external links to include and the recommended anchor text. Make sure they’re not competing articles.<br/> Split the outline into part 1 and part 2."
     },
     {
-        "id": "linkedin-ghostwriter",
+        "id": "linkedin-ghostwriter2",
         "name": "Act as Linkedin Ghostwriter",
         "value": "Act as an Expert Technical Architecture in Mobile, having more then 20 years of expertise in mobile technologies and development of various domain with cloud and native architecting design. Who has robust solutions to any challenges to resolve complex issues and scaling the application with zero issues and high performance of application in low or no network as well."
     }