neuralnode 2.2.8__tar.gz → 2.2.9__tar.gz

{neuralnode-2.2.8 → neuralnode-2.2.9}/.env.example

@@ -15,19 +15,6 @@ GOOGLE_API_KEY=your-google-api-key-here
 # No API key needed, just ensure Ollama is running
 OLLAMA_BASE_URL=http://localhost:11434
 
-# Horus Cyber Nano 1.0 BETA Early Access
-HORUS_CYBER_NANO_API_KEY=horusNN-your-access-token
-HORUS_CYBER_NANO_BASE_URL=https://api.tokenai.llc/v1
-
-# NeuralNode H-Preview API Gateway
-NEURALNODE_API_DB=neuralnode-api.sqlite3
-NEURALNODE_API_KEY=nn-your-24-character-token
-NEURALNODE_API_BASE_URL=http://127.0.0.1:8000/v1
-H_PREVIEW_UPSTREAM_URL=http://127.0.0.1:8001/v1
-H_PREVIEW_UPSTREAM_MODEL=DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU
-# Optional: only set this when the remote model server requires an API key.
-# H_PREVIEW_UPSTREAM_API_KEY=
-
 # NeuralNode Settings
 NEURALNODE_DEFAULT_PROVIDER=auto
 NEURALNODE_MEMORY_ENABLED=true

{neuralnode-2.2.8 → neuralnode-2.2.9}/.github/workflows/tests.yml

@@ -29,7 +29,7 @@ jobs:
     - name: Install wheel and verify package version
       run: |
         python -m pip install --force-reinstall .ci-dist/*.whl
-        python -c "import neuralnode; assert neuralnode.__version__ == '2.2.8'; print(neuralnode.__version__)"
+        python -c "import neuralnode; assert neuralnode.__version__ == '2.2.9'; print(neuralnode.__version__)"
 
   test:
     # Python 3.9 is no longer available on the newest Ubuntu runner image.

{neuralnode-2.2.8 → neuralnode-2.2.9}/.gitignore

@@ -6,6 +6,10 @@ __pycache__/
 *.egg-info/
 dist/
 release-dist/
+.release-dist*/
+.ci-dist/
+.ci-dist-verify*/
+.install-check*/
 build/
 .eggs/
 
@@ -23,7 +27,6 @@ env/
 !**/.env.example
 
 # Temp / outputs
-temp_header.gguf
 outputs/
 .tmp_*
 *.gguf.tmp

neuralnode-2.2.9/CHANGELOG.md

@@ -0,0 +1,89 @@
+# Changelog
+
+All notable changes to NeuralNode are documented here.
+
+## 2.2.9 — 2026-11-01
+
+### Integrations and runtime cleanup
+
+- Removed the deprecated WhatsApp/Twilio integration and legacy API gateway.
+- Removed obsolete provider and deployment artifacts from the maintained core.
+- Updated the OpenClaw manager for the current OpenClaw configuration schema,
+  loopback-only gateway binding, token authentication, and OpenAI-compatible
+  chat-completions enablement.
+- Added a blocking-chat fallback for OpenClaw streaming requests so providers
+  without native streaming can still complete gateway turns.
+- Validated the NeuralNode → OpenClaw → local Qwen/llama.cpp path on AWS.
+- Upgraded Ollama chat integration to the native `/api/chat` protocol with
+  role-preserving messages, streaming, structured output, and tool calls.
+- Added Ollama native reasoning control through `thinking=True` or
+  `thinking=False`, preserving the model trace in response metadata.
+- Updated Ollama embeddings to the batch `/api/embed` endpoint with a legacy
+  compatibility fallback.
+- Removed the remaining Twilio optional dependency with the retired WhatsApp
+  integration.
+
+## 2.2.8 — 2026-11-01
+
+NeuralNode 2.2.8 is a feature and reliability release focused on local model
+execution, multimodal workflows, integrations, and reproducible packaging.
+
+### Multimodal and local model support
+
+- Added local Hugging Face speech adapters:
+  - `HuggingFaceSTT` for audio transcription.
+  - `HuggingFaceTTS` for text-to-speech generation.
+- Added configurable Hugging Face model IDs, lazy loading, language handling,
+  CPU/CUDA device selection, and WAV output for the speech adapters.
+- Added local vision-language input support through the existing image chat
+  interface.
+- Added Diffusers-compatible local text-to-image loading and generation.
+- Improved local model generation determinism and provider identity.
+- Improved model repository download handling for Hugging Face sources.
+- Added explicit authentication support for private and gated Hugging Face
+  repositories using `HF_TOKEN` or an explicit token parameter.
+- Made optional Horus runtime imports tolerant of environments where the heavy
+  local ML dependencies are not installed.
+
+### Voice workflows
+
+- Added a sequential `VoiceAgent` pipeline for file-based and push-to-talk
+  interactions:
+
+  `audio input → speech-to-text → agent/LLM turn → text-to-speech → audio output`
+
+- Added HTTP speech adapters for connecting external STT and TTS services.
+- Documented the distinction between sequential voice turns and low-latency,
+  full-duplex live audio. Full-duplex streaming is outside this release.
+
+### Agents and tools
+
+- Improved provider-backed agent tool calling.
+- Added support for textual actions from tool-capable models.
+- Improved tool-result message handling and buffered streaming compatibility.
+- Preserved the existing agents, memory, RAG, embeddings, vector-store, and
+  integration APIs.
+
+### Horus and OpenClaw
+
+- Aligned Horus and OpenClaw integrations with the current public interfaces.
+- Secured the OpenClaw local API behavior.
+- Added test paths that do not require optional OpenClaw HTTP test packages.
+### Release engineering
+
+- Added source distribution and wheel build validation in CI.
+- Isolated package build output from repository working files.
+- Added trusted PyPI publishing through GitHub Actions and PyPI's OIDC
+  trusted-publishing flow.
+- Added public repository, ownership, license, and release metadata alignment.
+
+### Compatibility and limitations
+
+- Python 3.9, 3.10, 3.11, and 3.12 are supported by the package metadata.
+- Heavy local-model, image, and speech features remain optional extras; the
+  base installation does not install those large runtime dependencies.
+- Public Hugging Face models can be loaded without a token. Private or gated
+  repositories require a valid `HF_TOKEN`.
+- The voice pipeline in this release processes one turn at a time. It is not a
+  real-time, full-duplex audio session.
+

{neuralnode-2.2.8 → neuralnode-2.2.9}/PKG-INFO

@@ -1,14 +1,15 @@
 Metadata-Version: 2.5
 Name: neuralnode
-Version: 2.2.8
+Version: 2.2.9
 Summary: Connect models. Build AI. A Python framework for LLM providers, agents, memory, RAG, and tools.
-Project-URL: Homepage, https://assem.cloud/
+Project-URL: Homepage, https://tokenai.llc/
+Project-URL: NeuralNode, https://tokenai.llc/
 Project-URL: Documentation, https://neuralnode.readthedocs.io
 Project-URL: Repository, https://github.com/tokenaillc/neuralnode
 Project-URL: Issues, https://github.com/tokenaillc/neuralnode/issues
 Project-URL: Changelog, https://github.com/tokenaillc/neuralnode/blob/main/CHANGELOG.md
-Author-email: Assem Sabry <assemsabryy@outlook.com>
-Maintainer-email: Assem Sabry <assemsabryy@outlook.com>
+Author-email: Assem Sabry <assemsabry777@gmail.com>
+Maintainer-email: Assem Sabry <assemsabry777@gmail.com>
 License: TokenAI Source-Available License
         Version 1.0
         
@@ -217,13 +218,9 @@ Requires-Dist: speechrecognition>=3.10.0; extra == 'all'
 Requires-Dist: sqlalchemy>=2.0.0; extra == 'all'
 Requires-Dist: torch>=2.0.0; extra == 'all'
 Requires-Dist: transformers>=4.35.0; extra == 'all'
-Requires-Dist: twilio>=9.0.0; extra == 'all'
 Requires-Dist: uvicorn>=0.27.0; extra == 'all'
 Provides-Extra: anthropic
 Requires-Dist: anthropic>=0.8.0; extra == 'anthropic'
-Provides-Extra: api
-Requires-Dist: fastapi>=0.110.0; extra == 'api'
-Requires-Dist: uvicorn>=0.27.0; extra == 'api'
 Provides-Extra: aws
 Requires-Dist: boto3>=1.34.0; extra == 'aws'
 Provides-Extra: azure
@@ -351,12 +348,10 @@ Provides-Extra: vision
 Requires-Dist: pillow>=9.0; extra == 'vision'
 Provides-Extra: weaviate
 Requires-Dist: weaviate-client>=3.0.0; extra == 'weaviate'
-Provides-Extra: whatsapp
-Requires-Dist: twilio>=9.0.0; extra == 'whatsapp'
 Description-Content-Type: text/markdown
 
 <p align="center">
-    <img src="media/neuralnode.png" alt="NeuralNode — Connect models. Build AI." width="720">
+    <img src="https://raw.githubusercontent.com/tokenaillc/neuralnode/v2.2.8/media/neuralnode.png" alt="NeuralNode — Connect models. Build AI." width="720">
 </p>
 
 <p align="center">
@@ -375,6 +370,32 @@ one consistent interface. Build applications with model providers, agents,
 tools, memory, RAG, embeddings, vector stores, and integrations without
 rewriting your application around a single provider.
 
+## What's new in 2.2.8
+
+NeuralNode 2.2.8 expands the framework's local-model and multimodal building
+blocks while making the public release safer to build and publish.
+
+- Added Hugging Face authentication support for public, private, and gated
+  model repositories through the `HF_TOKEN` environment variable or an
+  explicit token argument.
+- Added local Hugging Face speech adapters for speech-to-text and text-to-speech,
+  including lazy loading, automatic CPU/CUDA selection, configurable model IDs,
+  language handling, and WAV output.
+- Added a sequential `VoiceAgent` workflow that connects audio transcription,
+  an agent turn, and speech synthesis for file-based and push-to-talk use cases.
+- Added local vision-language input and Diffusers-compatible text-to-image
+  generation through dedicated Hugging Face model loaders.
+- Improved Horus model loading, provider identity, deterministic generation,
+  optional runtime imports, and model repository download behavior.
+- Improved provider-backed agent tool calling, including textual actions and
+  tool-result handling.
+- Added and aligned OpenClaw integration behavior, including safer local API
+  access and test coverage that does not require optional HTTP test packages.
+- Hardened the release pipeline with source and wheel build validation,
+  isolated build output, and trusted publishing to PyPI through GitHub Actions.
+
+For the complete release record, see [CHANGELOG.md](CHANGELOG.md).
+
 ## Why NeuralNode?
 
 - **One interface for many models** — connect cloud APIs, local runtimes, and
@@ -506,49 +527,6 @@ environment. `HuggingFaceTTS.speak()` writes mono PCM WAV output.
 
 ## Horus
 
-### H-Preview Model API
-
-NeuralNode includes an OpenAI-compatible API gateway for serving H-Preview through a vLLM-compatible server on another machine. The gateway creates `nn-` access tokens with `Basic` (RPM controlled) and `Elite` (no application RPM accounting) tiers, plus one owner admin token for managing them. See [docs/model_api.md](docs/model_api.md) for deployment and Python client examples.
-
-```python
-import neuralnode as nn
-
-ai = nn.NeuralNode(
-    provider="h-preview",
-    model="h-preview",
-    api_key="nn-...",
-    base_url="https://api.example.com/v1",
-)
-print(ai.chat("Say hello."))
-```
-
-### Horus Cyber Nano 1.0 Early Access API
-
-Horus Cyber Nano 1.0 is a Text-to-Text model. The hosted API is currently in Early Access. Approved users can call it through the same
-`NeuralNode` interface. Each Access Token has its own server-enforced RPM limit,
-and API conversations are retained by TokenAI under the Early Access terms.
-
-```python
-import neuralnode as nn
-
-ai = nn.NeuralNode(
-    provider="horus-cyber-nano",
-    model="horus-cyber-nano-1.0-beta",
-    api_key="horusNN-XXXXXXXXXXXXXXXX",
-)
-
-print(ai.chat("Review this code for defensive cybersecurity issues."))
-```
-
-The client reads `HORUS_CYBER_NANO_API_KEY` automatically. Override
-`HORUS_CYBER_NANO_BASE_URL` only when TokenAI provides a different API endpoint.
-The production API base URL is `https://api.tokenai.llc/v1`. Agent Mode supports
-function tools, required or named tool choices, tool-result messages, and
-buffered SSE compatibility streams. The hosted Early Access model has a 32,768
-token context. See `examples/horus_cyber_nano_agent_mode.py` for the complete
-tool-call round trip. NeuralNode retries temporary HTTP 429, 502, and 503
-responses up to two times by default and respects the API Retry-After header.
-
 Locally loaded Horus models use:
 - unified chat template: `horus_unified`
 - unified context window: `8192`
@@ -766,7 +744,7 @@ If the `turboquant` package is not installed or the backend cannot use it, Neura
 
 ## Notes
 
-- OpenAI-compatible providers are fully supported (OpenAI, OpenRouter, AgentRouter, H-Preview, etc.).
+- OpenAI-compatible providers are fully supported (OpenAI, OpenRouter, and AgentRouter).
 - GGUF Horus models require `llama-cpp-python`.
 - Horus downloads from Hugging Face require `huggingface_hub`.
 
@@ -784,13 +762,12 @@ per-instance token automatically and configures its local server to require it.
 `send_task`; OpenClaw disables this endpoint by default. A stable `session_id` is sent
 as the OpenAI `user` value, so the Gateway creates or resumes that conversation when
 the first message is sent. OpenClaw session status/control uses its WebSocket RPC API
-and is not implemented by this HTTP client. Messaging channels such as WhatsApp must
-be configured and authorized in OpenClaw itself.
+and is not implemented by this HTTP client.
 
 ## Links
 
-- **TokenAI**: [https://tokenai.cloud/](https://tokenai.cloud/)
-- **NeuralNode**: [https://neuralnode.tokenai.cloud/](https://neuralnode.tokenai.cloud/)
+- **TokenAI**: [https://tokenai.llc/](https://tokenai.llc/)
+- **NeuralNode**: [https://tokenai.llc/](https://tokenai.llc/)
 
 ## License and ownership
 

{neuralnode-2.2.8 → neuralnode-2.2.9}/README.md

@@ -1,5 +1,5 @@
 <p align="center">
-    <img src="media/neuralnode.png" alt="NeuralNode — Connect models. Build AI." width="720">
+    <img src="https://raw.githubusercontent.com/tokenaillc/neuralnode/v2.2.8/media/neuralnode.png" alt="NeuralNode — Connect models. Build AI." width="720">
 </p>
 
 <p align="center">
@@ -18,6 +18,32 @@ one consistent interface. Build applications with model providers, agents,
 tools, memory, RAG, embeddings, vector stores, and integrations without
 rewriting your application around a single provider.
 
+## What's new in 2.2.8
+
+NeuralNode 2.2.8 expands the framework's local-model and multimodal building
+blocks while making the public release safer to build and publish.
+
+- Added Hugging Face authentication support for public, private, and gated
+  model repositories through the `HF_TOKEN` environment variable or an
+  explicit token argument.
+- Added local Hugging Face speech adapters for speech-to-text and text-to-speech,
+  including lazy loading, automatic CPU/CUDA selection, configurable model IDs,
+  language handling, and WAV output.
+- Added a sequential `VoiceAgent` workflow that connects audio transcription,
+  an agent turn, and speech synthesis for file-based and push-to-talk use cases.
+- Added local vision-language input and Diffusers-compatible text-to-image
+  generation through dedicated Hugging Face model loaders.
+- Improved Horus model loading, provider identity, deterministic generation,
+  optional runtime imports, and model repository download behavior.
+- Improved provider-backed agent tool calling, including textual actions and
+  tool-result handling.
+- Added and aligned OpenClaw integration behavior, including safer local API
+  access and test coverage that does not require optional HTTP test packages.
+- Hardened the release pipeline with source and wheel build validation,
+  isolated build output, and trusted publishing to PyPI through GitHub Actions.
+
+For the complete release record, see [CHANGELOG.md](CHANGELOG.md).
+
 ## Why NeuralNode?
 
 - **One interface for many models** — connect cloud APIs, local runtimes, and
@@ -149,49 +175,6 @@ environment. `HuggingFaceTTS.speak()` writes mono PCM WAV output.
 
 ## Horus
 
-### H-Preview Model API
-
-NeuralNode includes an OpenAI-compatible API gateway for serving H-Preview through a vLLM-compatible server on another machine. The gateway creates `nn-` access tokens with `Basic` (RPM controlled) and `Elite` (no application RPM accounting) tiers, plus one owner admin token for managing them. See [docs/model_api.md](docs/model_api.md) for deployment and Python client examples.
-
-```python
-import neuralnode as nn
-
-ai = nn.NeuralNode(
-    provider="h-preview",
-    model="h-preview",
-    api_key="nn-...",
-    base_url="https://api.example.com/v1",
-)
-print(ai.chat("Say hello."))
-```
-
-### Horus Cyber Nano 1.0 Early Access API
-
-Horus Cyber Nano 1.0 is a Text-to-Text model. The hosted API is currently in Early Access. Approved users can call it through the same
-`NeuralNode` interface. Each Access Token has its own server-enforced RPM limit,
-and API conversations are retained by TokenAI under the Early Access terms.
-
-```python
-import neuralnode as nn
-
-ai = nn.NeuralNode(
-    provider="horus-cyber-nano",
-    model="horus-cyber-nano-1.0-beta",
-    api_key="horusNN-XXXXXXXXXXXXXXXX",
-)
-
-print(ai.chat("Review this code for defensive cybersecurity issues."))
-```
-
-The client reads `HORUS_CYBER_NANO_API_KEY` automatically. Override
-`HORUS_CYBER_NANO_BASE_URL` only when TokenAI provides a different API endpoint.
-The production API base URL is `https://api.tokenai.llc/v1`. Agent Mode supports
-function tools, required or named tool choices, tool-result messages, and
-buffered SSE compatibility streams. The hosted Early Access model has a 32,768
-token context. See `examples/horus_cyber_nano_agent_mode.py` for the complete
-tool-call round trip. NeuralNode retries temporary HTTP 429, 502, and 503
-responses up to two times by default and respects the API Retry-After header.
-
 Locally loaded Horus models use:
 - unified chat template: `horus_unified`
 - unified context window: `8192`
@@ -409,7 +392,7 @@ If the `turboquant` package is not installed or the backend cannot use it, Neura
 
 ## Notes
 
-- OpenAI-compatible providers are fully supported (OpenAI, OpenRouter, AgentRouter, H-Preview, etc.).
+- OpenAI-compatible providers are fully supported (OpenAI, OpenRouter, and AgentRouter).
 - GGUF Horus models require `llama-cpp-python`.
 - Horus downloads from Hugging Face require `huggingface_hub`.
 
@@ -427,13 +410,12 @@ per-instance token automatically and configures its local server to require it.
 `send_task`; OpenClaw disables this endpoint by default. A stable `session_id` is sent
 as the OpenAI `user` value, so the Gateway creates or resumes that conversation when
 the first message is sent. OpenClaw session status/control uses its WebSocket RPC API
-and is not implemented by this HTTP client. Messaging channels such as WhatsApp must
-be configured and authorized in OpenClaw itself.
+and is not implemented by this HTTP client.
 
 ## Links
 
-- **TokenAI**: [https://tokenai.cloud/](https://tokenai.cloud/)
-- **NeuralNode**: [https://neuralnode.tokenai.cloud/](https://neuralnode.tokenai.cloud/)
+- **TokenAI**: [https://tokenai.llc/](https://tokenai.llc/)
+- **NeuralNode**: [https://tokenai.llc/](https://tokenai.llc/)
 
 ## License and ownership
 

{neuralnode-2.2.8 → neuralnode-2.2.9}/pyproject.toml

@@ -4,17 +4,17 @@ build-backend = "hatchling.build"
 
 [project]
 name = "neuralnode"
-version = "2.2.8"
+version = "2.2.9"
 description = "Connect models. Build AI. A Python framework for LLM providers, agents, memory, RAG, and tools."
 readme = "README.md"
 requires-python = ">=3.9"
 license = {file = "LICENSE"}
 keywords = ["llm", "ai", "agent", "framework", "openai", "anthropic", "google", "ollama", "langchain", "rag", "tools", "automation", "embeddings", "vector-stores"]
 authors = [
-    {name = "Assem Sabry", email = "assemsabryy@outlook.com"}
+    {name = "Assem Sabry", email = "assemsabry777@gmail.com"}
 ]
 maintainers = [
-    {name = "Assem Sabry", email = "assemsabryy@outlook.com"}
+    {name = "Assem Sabry", email = "assemsabry777@gmail.com"}
 ]
 classifiers = [
     "Development Status :: 4 - Beta",
@@ -169,15 +169,10 @@ telegram = [
 ]
 discord = ["discord.py>=2.3.0"]
 slack = ["slack-sdk>=3.21.0"]
-whatsapp = ["twilio>=9.0.0"]
 openclaw = [
     "fastapi>=0.110.0",
     "uvicorn>=0.27.0",
 ]
-api = [
-    "fastapi>=0.110.0",
-    "uvicorn>=0.27.0",
-]
 
 # All features
 all = [
@@ -202,7 +197,6 @@ all = [
     "einops>=0.7.0",
     "sentence-transformers>=2.2.0",
     "python-telegram-bot>=20.0",
-    "twilio>=9.0.0",
     "fastapi>=0.110.0",
     "uvicorn>=0.27.0",
     "edge-tts>=6.1.0",
@@ -228,15 +222,13 @@ dev = [
 ]
 
 [project.urls]
-Homepage = "https://assem.cloud/"
+Homepage = "https://tokenai.llc/"
+NeuralNode = "https://tokenai.llc/"
 Documentation = "https://neuralnode.readthedocs.io"
 Repository = "https://github.com/tokenaillc/neuralnode"
 Issues = "https://github.com/tokenaillc/neuralnode/issues"
 Changelog = "https://github.com/tokenaillc/neuralnode/blob/main/CHANGELOG.md"
 
-[project.scripts]
-neuralnode-api = "neuralnode.api:main"
-
 [tool.hatch.build.targets.wheel]
 packages = ["src/neuralnode"]
 exclude = [
@@ -251,11 +243,9 @@ exclude = [
     "/.tmp_*",
     "/actionlint.zip",
     "/actionlint_bin",
-    "/deployment",
     "/dist",
     "/dist_*",
     "/outputs",
-    "/temp_header.gguf",
     "**/__pycache__/**",
     "**/*.pyc",
 ]

{neuralnode-2.2.8 → neuralnode-2.2.9}/src/neuralnode/__init__.py

@@ -42,7 +42,7 @@ Quick Start::
     text = sr.listen()
 """
 
-__version__ = "2.2.8"
+__version__ = "2.2.9"
 __author__ = "NeuralNode Contributors"
 
 # ── Core types ────────────────────────────────────────────────────────────────
@@ -88,10 +88,6 @@ from .providers import (
     MistralProvider,
     CohereProvider,
     DeepSeekProvider,
-    HorusCyberNano,
-    HorusCyberNanoAPIError,
-    HorusCyberNanoProvider,
-    HPreviewProvider,
     LlamaCppProvider,
     TransformersProvider,
     UniversalLocalProvider,
@@ -263,16 +259,6 @@ try:
 except ImportError:
     _SLACK_AVAILABLE = False
 
-try:
-    from .integrations.whatsapp import (
-        TwilioWhatsAppBot,
-        create_twilio_whatsapp_bot,
-    )
-
-    _WHATSAPP_AVAILABLE = True
-except ImportError:
-    _WHATSAPP_AVAILABLE = False
-
 try:
     from .integrations.openclaw import (
         OpenClawManager,
@@ -287,14 +273,6 @@ except ImportError:
 # Logging helper
 from .utils.logger import get_logger, setup_logging
 from .utils.dependencies import ensure_feature_dependencies
-from .api import (
-    APIKeyStore,
-    H_PREVIEW_HF_REPO,
-    H_PREVIEW_MODEL,
-    create_api_app,
-    generate_admin_token,
-    run_api_server,
-)
 
 # ── Skill → tools mapping ─────────────────────────────────────────────────────
 
@@ -433,6 +411,8 @@ class NeuralNode:
         system_prompt: str = None,
         history: list = None,
         stream: bool = False,
+        thinking=None,
+        **kwargs,
     ):
         """Send a single message and get a response.
 
@@ -465,7 +445,14 @@ class NeuralNode:
                     messages.append(Message.assistant(content))
 
         messages.append(Message.user(message))
-        response = self._llm.chat(messages, stream=stream)
+        provider_kwargs = {
+            "temperature": self.temperature,
+            "max_tokens": self.max_tokens,
+            **kwargs,
+        }
+        if thinking is not None:
+            provider_kwargs["thinking"] = thinking
+        response = self._llm.chat(messages, stream=stream, **provider_kwargs)
 
         if stream:
             return (chunk.content for chunk in ensure_streaming_chunks(response))
@@ -737,16 +724,6 @@ __all__ = [
     "MistralProvider",
     "CohereProvider",
     "DeepSeekProvider",
-    "HorusCyberNano",
-    "HorusCyberNanoAPIError",
-    "HorusCyberNanoProvider",
-    "HPreviewProvider",
-    "APIKeyStore",
-    "H_PREVIEW_HF_REPO",
-    "H_PREVIEW_MODEL",
-    "create_api_app",
-    "generate_admin_token",
-    "run_api_server",
     "LlamaCppProvider",
     "TransformersProvider",
     "UniversalLocalProvider",
@@ -872,11 +849,6 @@ if _TURBOQUANT_AVAILABLE:
     __all__ += ["TurboQuant", "TurboQuantEngine", "TurboQuantConfig"]
 if _THINKING_AVAILABLE:
     __all__ += ["ThinkingMode", "ThinkingResponse", "enable_thinking"]
-if _WHATSAPP_AVAILABLE:
-    __all__ += [
-        "TwilioWhatsAppBot",
-        "create_twilio_whatsapp_bot",
-    ]
 if _OPENCLAW_AVAILABLE:
     __all__ += [
         "OpenClawManager",

{neuralnode-2.2.8 → neuralnode-2.2.9}/src/neuralnode/core/__init__.py

@@ -148,6 +148,13 @@ def ensure_llm_response(
                     )
                 )
 
+        metadata = dict(response.get("metadata", {}) or {})
+        # Preserve provider-specific reasoning traces (for example Ollama's
+        # native ``message.thinking``) without changing the stable response
+        # shape used by existing callers.
+        if response.get("thinking") is not None and "thinking" not in metadata:
+            metadata["thinking"] = response.get("thinking")
+
         return LLMResponse(
             content=response.get("content", ""),
             model=response.get("model", default_model),
@@ -155,7 +162,7 @@ def ensure_llm_response(
             tool_calls=tool_calls,
             usage=response.get("usage", {}) or {},
             finish_reason=response.get("finish_reason"),
-            metadata=response.get("metadata", {}) or {},
+            metadata=metadata,
         )
 
     raise TypeError(f"Unsupported response type: {type(response)!r}")

{neuralnode-2.2.8 → neuralnode-2.2.9}/src/neuralnode/integrations/__init__.py

@@ -9,7 +9,6 @@ from .telegram import (
 )
 from .discord import DiscordBot
 from .slack import SlackBot
-from .whatsapp import TwilioWhatsAppBot, create_twilio_whatsapp_bot
 from .openclaw import OpenClawManager, OpenClawServer, OpenClawClient
 
 __all__ = [
@@ -20,8 +19,6 @@ __all__ = [
     "create_telegram_bot",
     "DiscordBot",
     "SlackBot",
-    "TwilioWhatsAppBot",
-    "create_twilio_whatsapp_bot",
     "OpenClawManager",
     "OpenClawServer",
     "OpenClawClient",