llms-py 3.0.0b9__tar.gz → 3.0.0b10__tar.gz

{llms_py-3.0.0b9/llms_py.egg-info → llms_py-3.0.0b10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llms-py
-Version: 3.0.0b9
+Version: 3.0.0b10
 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-3.0.0b10/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-3.0.0b10}/llms/extensions/app/__init__.py

@@ -60,7 +60,7 @@ 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 +174,13 @@ 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 +190,8 @@ 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 +343,22 @@ 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-3.0.0b10}/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-3.0.0b10}/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-3.0.0b10}/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-3.0.0b10/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-3.0.0b10}/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-3.0.0b10}/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-3.0.0b10/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-3.0.0b10}/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-3.0.0b10/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-3.0.0b10/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-3.0.0b10}/llms/llms.json

@@ -9,14 +9,11 @@
             "restrict_to": "GITHUB_USERS"
         }
     },
-    "disable_extensions": [
-        "xmas",
-        "duckduckgo"
-    ],
+    "disable_extensions": [],
     "defaults": {
         "headers": {
             "Content-Type": "application/json",
-            "User-Agent": "llmspy.org/1.0"
+            "User-Agent": "llmspy.org/3.0"
         },
         "text": {
             "model": "kimi-k2",
@@ -95,7 +92,7 @@
             ]
         },
         "out:image": {
-            "model": "black-forest-labs/flux.1-dev",
+            "model": "gemini-2.5-flash-image",
             "messages": [
                 {
                     "role": "user",
@@ -202,10 +199,10 @@
             }
         },
         "github-copilot": {
-            "enabled": false
+            "enabled": true
         },
         "github-models": {
-            "enabled": false,
+            "enabled": true,
             "check": {
                 "messages": [
                     {
@@ -230,13 +227,13 @@
             "temperature": 1.0
         },
         "ollama": {
-            "enabled": true,
+            "enabled": false,
             "id": "ollama",
             "npm": "ollama",
             "api": "http://localhost:11434"
         },
         "lmstudio": {
-            "enabled": true,
+            "enabled": false,
             "npm": "lmstudio",
             "api": "http://127.0.0.1:1234/v1",
             "models": {}
@@ -351,7 +348,7 @@
             "enabled": true
         },
         "moonshotai": {
-            "enabled": false
+            "enabled": true
         },
         "nvidia": {
             "enabled": true,
@@ -372,7 +369,7 @@
             "enabled": true
         },
         "fireworks-ai": {
-            "enabled": false
+            "enabled": true
         },
         "openrouter": {
             "enabled": true,

{llms_py-3.0.0b9 → llms_py-3.0.0b10}/llms/main.py

@@ -29,7 +29,7 @@ from importlib import resources  # Py≥3.9  (pip install importlib_resources fo
 from io import BytesIO
 from pathlib import Path
 from typing import get_type_hints
-from urllib.parse import parse_qs, urlencode
+from urllib.parse import parse_qs, urlencode, urljoin
 
 import aiohttp
 from aiohttp import web
@@ -41,7 +41,7 @@ try:
 except ImportError:
     HAS_PIL = False
 
-VERSION = "3.0.0b9"
+VERSION = "3.0.0b10"
 _ROOT = None
 DEBUG = os.getenv("DEBUG") == "1"
 MOCK = os.getenv("MOCK") == "1"
@@ -204,6 +204,12 @@ def id_to_name(id):
     return id.replace("-", " ").title()
 
 
+def pluralize(word, count):
+    if count == 1:
+        return word
+    return word + "s"
+
+
 def get_file_mime_type(filename):
     mime_type, _ = mimetypes.guess_type(filename)
     return mime_type or "application/octet-stream"
@@ -368,6 +374,9 @@ async def process_chat(chat, provider_id=None):
         raise Exception("No chat provided")
     if "stream" not in chat:
         chat["stream"] = False
+    # Some providers don't support empty tools
+    if "tools" in chat and len(chat["tools"]) == 0:
+        del chat["tools"]
     if "messages" not in chat:
         return chat
 
@@ -700,6 +709,7 @@ def save_image_to_cache(base64_data, filename, image_info, ignore_info=False):
 async def response_json(response):
     text = await response.text()
     if response.status >= 400:
+        _dbg(f"HTTP {response.status} {response.reason}: {text}")
         raise HTTPError(response.status, reason=response.reason, body=text, headers=dict(response.headers))
     response.raise_for_status()
     body = json.loads(text)
@@ -1427,6 +1437,8 @@ async def g_chat_completion(chat, context=None):
                     current_chat["messages"].append(message)
                     tool_history.append(message)
 
+                    await g_app.on_chat_tool(current_chat, context)
+
                     for tool_call in tool_calls:
                         function_name = tool_call["function"]["name"]
                         try:
@@ -1450,8 +1462,7 @@ async def g_chat_completion(chat, context=None):
                         current_chat["messages"].append(tool_msg)
                         tool_history.append(tool_msg)
 
-                    for filter_func in g_app.chat_tool_filters:
-                        await filter_func(current_chat, context)
+                    await g_app.on_chat_tool(current_chat, context)
 
                     if should_cancel_thread(context):
                         return
@@ -1606,7 +1617,7 @@ async def cli_chat(chat, tools=None, image=None, audio=None, file=None, args=Non
                 for file in generated_files:
                     if file.startswith("/~cache"):
                         print(get_cache_path(file[8:]))
-                        print(f"http://localhost:8000/{file}")
+                        print(urljoin("http://localhost:8000", file))
                     else:
                         print(file)
 
@@ -2443,6 +2454,15 @@ class AppExtensions:
             except Exception as e:
                 _err("chat error filter failed", e)
 
+    async def on_chat_tool(self, chat, context):
+        m_len = len(chat.get("messages", []))
+        t_len = len(self.chat_tool_filters)
+        _dbg(
+            f"on_tool_call for thread {context.get('threadId', None)} with {m_len} {pluralize('message', m_len)}, invoking {t_len} {pluralize('filter', t_len)}:"
+        )
+        for filter_func in self.chat_tool_filters:
+            await filter_func(chat, context)
+
     def exit(self, exit_code=0):
         if len(self.shutdown_handlers) > 0:
             _dbg(f"running {len(self.shutdown_handlers)} shutdown handlers...")

{llms_py-3.0.0b9 → llms_py-3.0.0b10}/llms/ui/ai.mjs

@@ -6,7 +6,7 @@ const headers = { 'Accept': 'application/json' }
 const prefsKey = 'llms.prefs'
 
 export const o = {
-    version: '3.0.0b9',
+    version: '3.0.0b10',
     base,
     prefsKey,
     welcome: 'Welcome to llms.py',
@@ -74,12 +74,27 @@ export const o = {
             return { responseStatus }
         }
     },
+    createErrorStatus({ message, errorCode, stackTrace, errors, meta }) {
+        const ret = {
+            errorCode: errorCode || 'Error',
+            message: message,
+        }
+        if (stackTrace) {
+            ret.stackTrace = stackTrace
+        }
+        if (errors && Array.isArray(errors)) {
+            ret.errors = errors
+        }
+        if (meta) {
+            ret.meta = meta
+        }
+        return ret
+    },
     createErrorResult(e) {
         return new ApiResult({
-            responseStatus: {
-                errorCode: 'Error',
-                message: `${e.message ?? e}`
-            }
+            error: e.errorCode
+                ? this.createErrorStatus(e)
+                : this.createErrorStatus({ message: `${e.message ?? e}` })
         })
     },
     async getConfig() {