agent-cli 0.73.0__tar.gz → 0.75.0__tar.gz

{agent_cli-0.73.0 → agent_cli-0.75.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-cli
-Version: 0.73.0
+Version: 0.75.0
 Summary: A suite of AI-powered command-line tools for text correction, audio transcription, and voice assistance.
 Project-URL: Homepage, https://github.com/basnijholt/agent-cli
 Author-email: Bas Nijholt <bas@nijho.lt>
@@ -46,8 +46,6 @@ Requires-Dist: soundfile>=0.12.0; extra == 'kokoro'
 Requires-Dist: transformers>=4.40.0; extra == 'kokoro'
 Provides-Extra: llm
 Requires-Dist: pydantic-ai-slim[duckduckgo,google,openai,vertexai]>=0.1.1; extra == 'llm'
-Provides-Extra: llm-core
-Requires-Dist: pydantic-ai-slim[google,openai]>=0.1.1; extra == 'llm-core'
 Provides-Extra: memory
 Requires-Dist: chromadb>=0.4.22; extra == 'memory'
 Requires-Dist: fastapi[standard]; extra == 'memory'
@@ -86,6 +84,10 @@ Requires-Dist: pytest-timeout; extra == 'test'
 Requires-Dist: pytest>=7.0.0; extra == 'test'
 Provides-Extra: vad
 Requires-Dist: silero-vad-lite>=0.2.1; extra == 'vad'
+Provides-Extra: whisper-transformers
+Requires-Dist: fastapi[standard]; extra == 'whisper-transformers'
+Requires-Dist: torch>=2.0.0; extra == 'whisper-transformers'
+Requires-Dist: transformers>=4.30.0; extra == 'whisper-transformers'
 Provides-Extra: wyoming
 Requires-Dist: wyoming>=1.5.2; extra == 'wyoming'
 Description-Content-Type: text/markdown
@@ -426,7 +428,7 @@ Our installation scripts automatically handle all dependencies:
 |---------|---------|-----------------|
 | **[Ollama](https://ollama.ai/)** | Local LLM for text processing | ✅ Yes, with default model |
 | **[Wyoming Faster Whisper](https://github.com/rhasspy/wyoming-faster-whisper)** | Speech-to-text | ✅ Yes, via `uvx` |
-| **[`agent-cli server whisper`](docs/commands/server/whisper.md)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[whisper]"` |
+| **[`agent-cli server whisper`](docs/commands/server/whisper.md)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[faster-whisper]"` |
 | **[Wyoming Piper](https://github.com/rhasspy/wyoming-piper)** | Text-to-speech | ✅ Yes, via `uvx` |
 | **[Kokoro-FastAPI](https://github.com/remsky/Kokoro-FastAPI)** | Premium TTS (optional) | ⚙️ Can be added later |
 | **[Wyoming openWakeWord](https://github.com/rhasspy/wyoming-openwakeword)** | Wake word detection | ✅ Yes, for `assistant` |
@@ -520,6 +522,7 @@ agent-cli install-extras rag memory vad
   • kokoro - Kokoro neural TTS engine
   • faster-whisper - Whisper ASR for CUDA/CPU
   • mlx-whisper - Whisper ASR for Apple Silicon
+  • whisper-transformers - Whisper ASR via HuggingFace transformers (safetensors)
   • wyoming - Wyoming protocol for ASR/TTS servers
   • server - FastAPI server components
   • speed - Audio speed adjustment
@@ -536,7 +539,8 @@ agent-cli install-extras rag memory vad
 
 ╭─ Arguments ────────────────────────────────────────────────────────────────────────────╮
 │   extras      [EXTRAS]...  Extras to install: rag, memory, vad, audio, piper, kokoro,  │
-│                            faster-whisper, mlx-whisper, wyoming, server, speed, llm    │
+│                            faster-whisper, mlx-whisper, whisper-transformers, wyoming, │
+│                            server, speed, llm                                          │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ Options ──────────────────────────────────────────────────────────────────────────────╮
 │ --list  -l        Show available extras with descriptions (what each one enables)      │
@@ -632,8 +636,8 @@ the `[defaults]` section of your configuration file.
 ```toml
 [defaults]
 # llm_provider = "ollama"  # 'ollama', 'openai', or 'gemini'
-# asr_provider = "wyoming" # 'wyoming' or 'openai'
-# tts_provider = "wyoming" # 'wyoming', 'openai', or 'kokoro'
+# asr_provider = "wyoming" # 'wyoming', 'openai', or 'gemini'
+# tts_provider = "wyoming" # 'wyoming', 'openai', 'kokoro', or 'gemini'
 # openai_api_key = "sk-..."
 # gemini_api_key = "..."
 ```
@@ -645,7 +649,7 @@ the `[defaults]` section of your configuration file.
 **Workflow:** This is a simple, one-shot command.
 
 1.  It reads text from your system clipboard (or from a direct argument).
-2.  It sends the text to a local Ollama LLM with a prompt to perform only technical corrections.
+2.  It sends the text to your configured LLM provider (default: Ollama) with a prompt to perform only technical corrections.
 3.  The corrected text is copied back to your clipboard, replacing the original.
 
 **How to Use It:** This tool is ideal for integrating with a system-wide hotkey.

{agent_cli-0.73.0 → agent_cli-0.75.0}/README.md

@@ -334,7 +334,7 @@ Our installation scripts automatically handle all dependencies:
 |---------|---------|-----------------|
 | **[Ollama](https://ollama.ai/)** | Local LLM for text processing | ✅ Yes, with default model |
 | **[Wyoming Faster Whisper](https://github.com/rhasspy/wyoming-faster-whisper)** | Speech-to-text | ✅ Yes, via `uvx` |
-| **[`agent-cli server whisper`](docs/commands/server/whisper.md)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[whisper]"` |
+| **[`agent-cli server whisper`](docs/commands/server/whisper.md)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[faster-whisper]"` |
 | **[Wyoming Piper](https://github.com/rhasspy/wyoming-piper)** | Text-to-speech | ✅ Yes, via `uvx` |
 | **[Kokoro-FastAPI](https://github.com/remsky/Kokoro-FastAPI)** | Premium TTS (optional) | ⚙️ Can be added later |
 | **[Wyoming openWakeWord](https://github.com/rhasspy/wyoming-openwakeword)** | Wake word detection | ✅ Yes, for `assistant` |
@@ -428,6 +428,7 @@ agent-cli install-extras rag memory vad
   • kokoro - Kokoro neural TTS engine
   • faster-whisper - Whisper ASR for CUDA/CPU
   • mlx-whisper - Whisper ASR for Apple Silicon
+  • whisper-transformers - Whisper ASR via HuggingFace transformers (safetensors)
   • wyoming - Wyoming protocol for ASR/TTS servers
   • server - FastAPI server components
   • speed - Audio speed adjustment
@@ -444,7 +445,8 @@ agent-cli install-extras rag memory vad
 
 ╭─ Arguments ────────────────────────────────────────────────────────────────────────────╮
 │   extras      [EXTRAS]...  Extras to install: rag, memory, vad, audio, piper, kokoro,  │
-│                            faster-whisper, mlx-whisper, wyoming, server, speed, llm    │
+│                            faster-whisper, mlx-whisper, whisper-transformers, wyoming, │
+│                            server, speed, llm                                          │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ Options ──────────────────────────────────────────────────────────────────────────────╮
 │ --list  -l        Show available extras with descriptions (what each one enables)      │
@@ -540,8 +542,8 @@ the `[defaults]` section of your configuration file.
 ```toml
 [defaults]
 # llm_provider = "ollama"  # 'ollama', 'openai', or 'gemini'
-# asr_provider = "wyoming" # 'wyoming' or 'openai'
-# tts_provider = "wyoming" # 'wyoming', 'openai', or 'kokoro'
+# asr_provider = "wyoming" # 'wyoming', 'openai', or 'gemini'
+# tts_provider = "wyoming" # 'wyoming', 'openai', 'kokoro', or 'gemini'
 # openai_api_key = "sk-..."
 # gemini_api_key = "..."
 ```
@@ -553,7 +555,7 @@ the `[defaults]` section of your configuration file.
 **Workflow:** This is a simple, one-shot command.
 
 1.  It reads text from your system clipboard (or from a direct argument).
-2.  It sends the text to a local Ollama LLM with a prompt to perform only technical corrections.
+2.  It sends the text to your configured LLM provider (default: Ollama) with a prompt to perform only technical corrections.
 3.  The corrected text is copied back to your clipboard, replacing the original.
 
 **How to Use It:** This tool is ideal for integrating with a system-wide hotkey.

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/_extras.json

@@ -6,21 +6,19 @@
     ]
   ],
   "faster-whisper": [
-    "TODO: add description",
-    []
+    "Whisper ASR via CTranslate2",
+    [
+      "faster_whisper"
+    ]
   ],
   "kokoro": [
-    "TODO: add description",
-    []
-  ],
-  "llm": [
-    "LLM framework (pydantic-ai)",
+    "Kokoro neural TTS (GPU)",
     [
-      "pydantic_ai"
+      "kokoro"
     ]
   ],
-  "llm-core": [
-    "Minimal LLM support for Docker images",
+  "llm": [
+    "LLM framework (pydantic-ai)",
     [
       "pydantic_ai"
     ]
@@ -34,12 +32,16 @@
     ]
   ],
   "mlx-whisper": [
-    "TODO: add description",
-    []
+    "Whisper ASR for Apple Silicon",
+    [
+      "mlx_whisper"
+    ]
   ],
   "piper": [
-    "TODO: add description",
-    []
+    "Piper TTS (CPU)",
+    [
+      "piper"
+    ]
   ],
   "rag": [
     "RAG proxy (ChromaDB, embeddings)",
@@ -66,6 +68,13 @@
       "silero_vad"
     ]
   ],
+  "whisper-transformers": [
+    "Whisper ASR via HuggingFace transformers",
+    [
+      "transformers",
+      "torch"
+    ]
+  ],
   "wyoming": [
     "Wyoming protocol support",
     [

agent_cli-0.75.0/agent_cli/_requirements/whisper-transformers.txt

@@ -0,0 +1,256 @@
+# This file was autogenerated by uv via the following command:
+#    uv export --extra whisper-transformers --no-dev --no-emit-project --no-hashes
+annotated-doc==0.0.4
+    # via fastapi
+annotated-types==0.7.0
+    # via pydantic
+anyio==4.12.1
+    # via
+    #   httpx
+    #   starlette
+    #   watchfiles
+certifi==2026.1.4
+    # via
+    #   httpcore
+    #   httpx
+    #   requests
+    #   sentry-sdk
+charset-normalizer==3.4.4
+    # via requests
+click==8.3.1
+    # via
+    #   rich-toolkit
+    #   typer
+    #   typer-slim
+    #   uvicorn
+colorama==0.4.6 ; sys_platform == 'win32'
+    # via
+    #   click
+    #   tqdm
+    #   uvicorn
+dnspython==2.8.0
+    # via email-validator
+dotenv==0.9.9
+    # via agent-cli
+email-validator==2.3.0
+    # via
+    #   fastapi
+    #   pydantic
+fastapi==0.128.0
+    # via agent-cli
+fastapi-cli==0.0.20
+    # via fastapi
+fastapi-cloud-cli==0.10.1
+    # via fastapi-cli
+fastar==0.8.0
+    # via fastapi-cloud-cli
+filelock==3.20.3
+    # via
+    #   huggingface-hub
+    #   torch
+    #   transformers
+fsspec==2026.1.0
+    # via
+    #   huggingface-hub
+    #   torch
+h11==0.16.0
+    # via
+    #   httpcore
+    #   uvicorn
+hf-xet==1.2.0 ; platform_machine == 'aarch64' or platform_machine == 'amd64' or platform_machine == 'arm64' or platform_machine == 'x86_64'
+    # via huggingface-hub
+httpcore==1.0.9
+    # via httpx
+httptools==0.7.1
+    # via uvicorn
+httpx==0.28.1
+    # via
+    #   agent-cli
+    #   fastapi
+    #   fastapi-cloud-cli
+huggingface-hub==0.36.0
+    # via
+    #   tokenizers
+    #   transformers
+idna==3.11
+    # via
+    #   anyio
+    #   email-validator
+    #   httpx
+    #   requests
+jinja2==3.1.6
+    # via
+    #   fastapi
+    #   torch
+markdown-it-py==4.0.0
+    # via rich
+markupsafe==3.0.3
+    # via jinja2
+mdurl==0.1.2
+    # via markdown-it-py
+mpmath==1.3.0
+    # via sympy
+networkx==3.6.1
+    # via torch
+numpy==2.3.5
+    # via transformers
+nvidia-cublas-cu12==12.8.4.1 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via
+    #   nvidia-cudnn-cu12
+    #   nvidia-cusolver-cu12
+    #   torch
+nvidia-cuda-cupti-cu12==12.8.90 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cuda-nvrtc-cu12==12.8.93 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cuda-runtime-cu12==12.8.90 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cudnn-cu12==9.10.2.21 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cufft-cu12==11.3.3.83 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cufile-cu12==1.13.1.3 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-curand-cu12==10.3.9.90 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cusolver-cu12==11.7.3.90 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-cusparse-cu12==12.5.8.93 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via
+    #   nvidia-cusolver-cu12
+    #   torch
+nvidia-cusparselt-cu12==0.7.1 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-nccl-cu12==2.27.5 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-nvjitlink-cu12==12.8.93 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via
+    #   nvidia-cufft-cu12
+    #   nvidia-cusolver-cu12
+    #   nvidia-cusparse-cu12
+    #   torch
+nvidia-nvshmem-cu12==3.3.20 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+nvidia-nvtx-cu12==12.8.90 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+packaging==25.0
+    # via
+    #   huggingface-hub
+    #   transformers
+psutil==7.2.1 ; sys_platform == 'win32'
+    # via agent-cli
+pydantic==2.12.5
+    # via
+    #   agent-cli
+    #   fastapi
+    #   fastapi-cloud-cli
+    #   pydantic-extra-types
+    #   pydantic-settings
+pydantic-core==2.41.5
+    # via pydantic
+pydantic-extra-types==2.11.0
+    # via fastapi
+pydantic-settings==2.12.0
+    # via fastapi
+pygments==2.19.2
+    # via rich
+pyperclip==1.11.0
+    # via agent-cli
+python-dotenv==1.2.1
+    # via
+    #   dotenv
+    #   pydantic-settings
+    #   uvicorn
+python-multipart==0.0.21
+    # via fastapi
+pyyaml==6.0.3
+    # via
+    #   huggingface-hub
+    #   transformers
+    #   uvicorn
+regex==2026.1.15
+    # via transformers
+requests==2.32.5
+    # via
+    #   huggingface-hub
+    #   transformers
+rich==14.2.0
+    # via
+    #   agent-cli
+    #   rich-toolkit
+    #   typer
+    #   typer-slim
+rich-toolkit==0.17.1
+    # via
+    #   fastapi-cli
+    #   fastapi-cloud-cli
+rignore==0.7.6
+    # via fastapi-cloud-cli
+safetensors==0.7.0
+    # via transformers
+sentry-sdk==2.49.0
+    # via fastapi-cloud-cli
+setproctitle==1.3.7
+    # via agent-cli
+setuptools==80.9.0 ; python_full_version >= '3.12'
+    # via torch
+shellingham==1.5.4
+    # via
+    #   typer
+    #   typer-slim
+starlette==0.50.0
+    # via fastapi
+sympy==1.14.0
+    # via torch
+tokenizers==0.22.2
+    # via transformers
+torch==2.9.1
+    # via agent-cli
+tqdm==4.67.1
+    # via
+    #   huggingface-hub
+    #   transformers
+transformers==4.57.5
+    # via agent-cli
+triton==3.5.1 ; platform_machine == 'x86_64' and sys_platform == 'linux'
+    # via torch
+typer==0.21.1
+    # via
+    #   agent-cli
+    #   fastapi-cli
+    #   fastapi-cloud-cli
+typer-slim==0.21.1
+    # via agent-cli
+typing-extensions==4.15.0
+    # via
+    #   anyio
+    #   fastapi
+    #   huggingface-hub
+    #   pydantic
+    #   pydantic-core
+    #   pydantic-extra-types
+    #   rich-toolkit
+    #   starlette
+    #   torch
+    #   typer
+    #   typer-slim
+    #   typing-inspection
+typing-inspection==0.4.2
+    # via
+    #   pydantic
+    #   pydantic-settings
+urllib3==2.3.0
+    # via
+    #   requests
+    #   sentry-sdk
+uvicorn==0.40.0
+    # via
+    #   fastapi
+    #   fastapi-cli
+    #   fastapi-cloud-cli
+uvloop==0.22.1 ; platform_python_implementation != 'PyPy' and sys_platform != 'cygwin' and sys_platform != 'win32'
+    # via uvicorn
+watchfiles==1.1.1
+    # via uvicorn
+websockets==15.0.1
+    # via uvicorn

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/example-config.toml

@@ -13,7 +13,10 @@
 [defaults]
 
 # --- Provider Selection ---
-# Select the default provider for each service ("ollama"/"openai"/"gemini" for LLM, "wyoming"/"openai" for ASR).
+# Select the default provider for each service.
+# LLM: "ollama", "openai", or "gemini"
+# ASR: "wyoming", "openai", or "gemini"
+# TTS: "wyoming", "openai", "kokoro", or "gemini"
 llm-provider = "ollama"  # "local" still works as a deprecated alias
 tts-provider = "wyoming"
 

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/install/extras.py

@@ -63,7 +63,9 @@ def _install_via_uv_tool(extras: list[str], *, quiet: bool = False) -> bool:
     """Reinstall agent-cli via uv tool with the specified extras."""
     extras_str = ",".join(extras)
     package_spec = f"agent-cli[{extras_str}]"
-    python_version = f"{sys.version_info.major}.{sys.version_info.minor}"
+    # Cap at Python 3.13 for compatibility - some deps (e.g., silero-vad-lite) don't support 3.14 yet
+    major, minor = sys.version_info[:2]
+    python_version = f"{major}.{min(minor, 13)}"
     cmd = ["uv", "tool", "install", package_spec, "--force", "--python", python_version]
     if quiet:
         cmd.append("-q")
@@ -130,7 +132,8 @@ def install_extras(
         list[str] | None,
         typer.Argument(
             help="Extras to install: `rag`, `memory`, `vad`, `audio`, `piper`, `kokoro`, "
-            "`faster-whisper`, `mlx-whisper`, `wyoming`, `server`, `speed`, `llm`",
+            "`faster-whisper`, `mlx-whisper`, `whisper-transformers`, `wyoming`, `server`, "
+            "`speed`, `llm`",
         ),
     ] = None,
     list_extras: Annotated[
@@ -161,6 +164,7 @@ def install_extras(
     - `kokoro` - Kokoro neural TTS engine
     - `faster-whisper` - Whisper ASR for CUDA/CPU
     - `mlx-whisper` - Whisper ASR for Apple Silicon
+    - `whisper-transformers` - Whisper ASR via HuggingFace transformers (safetensors)
     - `wyoming` - Wyoming protocol for ASR/TTS servers
     - `server` - FastAPI server components
     - `speed` - Audio speed adjustment

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/scripts/sync_extras.py

@@ -33,7 +33,6 @@ EXTRA_METADATA: dict[str, tuple[str, list[str]]] = {
     "openai": ("OpenAI API provider", ["openai"]),
     "gemini": ("Google Gemini provider", ["google.genai"]),
     "llm": ("LLM framework (pydantic-ai)", ["pydantic_ai"]),
-    "llm-core": ("Minimal LLM support for Docker images", ["pydantic_ai"]),
     # Feature extras
     "rag": ("RAG proxy (ChromaDB, embeddings)", ["chromadb", "pydantic_ai"]),
     "memory": ("Long-term memory proxy", ["chromadb", "yaml", "pydantic_ai"]),
@@ -44,6 +43,16 @@ EXTRA_METADATA: dict[str, tuple[str, list[str]]] = {
     "tts-kokoro": ("Kokoro neural TTS", ["kokoro"]),
     "server": ("FastAPI server components", ["fastapi"]),
     "speed": ("Audio speed adjustment (audiostretchy)", ["audiostretchy"]),
+    # Whisper backends
+    "faster-whisper": ("Whisper ASR via CTranslate2", ["faster_whisper"]),
+    "mlx-whisper": ("Whisper ASR for Apple Silicon", ["mlx_whisper"]),
+    "whisper-transformers": (
+        "Whisper ASR via HuggingFace transformers",
+        ["transformers", "torch"],
+    ),
+    # TTS backends
+    "piper": ("Piper TTS (CPU)", ["piper"]),
+    "kokoro": ("Kokoro neural TTS (GPU)", ["kokoro"]),
 }
 
 

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/server/cli.py

@@ -157,36 +157,43 @@ def _download_tts_models(
 def _check_whisper_deps(backend: str, *, download_only: bool = False) -> None:
     """Check that Whisper dependencies are available."""
     _check_server_deps()
-    if download_only:
-        if not _has("faster_whisper"):
+    if backend == "mlx":
+        if not _has("mlx_whisper"):
             err_console.print(
-                "[bold red]Error:[/bold red] faster-whisper is required for --download-only. "
-                "Run: [cyan]pip install agent-cli\\[whisper][/cyan] "
-                "or [cyan]uv sync --extra whisper[/cyan]",
+                "[bold red]Error:[/bold red] MLX Whisper backend requires mlx-whisper. "
+                "Run: [cyan]pip install mlx-whisper[/cyan]",
             )
             raise typer.Exit(1)
         return
 
-    if backend == "mlx":
-        if not _has("mlx_whisper"):
+    if backend == "transformers":
+        if not _has("transformers") or not _has("torch"):
             err_console.print(
-                "[bold red]Error:[/bold red] MLX Whisper backend requires mlx-whisper. "
-                "Run: [cyan]pip install mlx-whisper[/cyan]",
+                "[bold red]Error:[/bold red] Transformers backend requires transformers and torch. "
+                "Run: [cyan]pip install agent-cli\\[whisper-transformers][/cyan] "
+                "or [cyan]uv sync --extra whisper-transformers[/cyan]",
             )
             raise typer.Exit(1)
         return
 
     if not _has("faster_whisper"):
-        err_console.print(
-            "[bold red]Error:[/bold red] Whisper dependencies not installed. "
-            "Run: [cyan]pip install agent-cli\\[whisper][/cyan] "
-            "or [cyan]uv sync --extra whisper[/cyan]",
-        )
+        if download_only:
+            err_console.print(
+                "[bold red]Error:[/bold red] faster-whisper is required for --download-only. "
+                "Run: [cyan]pip install agent-cli\\[faster-whisper][/cyan] "
+                "or [cyan]uv sync --extra faster-whisper[/cyan]",
+            )
+        else:
+            err_console.print(
+                "[bold red]Error:[/bold red] Whisper dependencies not installed. "
+                "Run: [cyan]pip install agent-cli\\[faster-whisper][/cyan] "
+                "or [cyan]uv sync --extra faster-whisper[/cyan]",
+            )
         raise typer.Exit(1)
 
 
 @app.command("whisper")
-@requires_extras("server", "faster-whisper|mlx-whisper", "wyoming")
+@requires_extras("server", "faster-whisper|mlx-whisper|whisper-transformers", "wyoming")
 def whisper_cmd(  # noqa: PLR0912, PLR0915
     model: Annotated[
         list[str] | None,
@@ -299,7 +306,7 @@ def whisper_cmd(  # noqa: PLR0912, PLR0915
             "-b",
             help=(
                 "Inference backend: `auto` (faster-whisper on CUDA/CPU, MLX on Apple Silicon), "
-                "`faster-whisper`, `mlx`"
+                "`faster-whisper`, `mlx`, `transformers` (HuggingFace, supports safetensors)"
             ),
         ),
     ] = "auto",
@@ -331,7 +338,7 @@ def whisper_cmd(  # noqa: PLR0912, PLR0915
     # Setup Rich logging for consistent output
     setup_rich_logging(log_level)
 
-    valid_backends = ("auto", "faster-whisper", "mlx")
+    valid_backends = ("auto", "faster-whisper", "mlx", "transformers")
     if backend not in valid_backends:
         err_console.print(
             f"[bold red]Error:[/bold red] --backend must be one of: {', '.join(valid_backends)}",
@@ -339,7 +346,7 @@ def whisper_cmd(  # noqa: PLR0912, PLR0915
         raise typer.Exit(1)
 
     resolved_backend = backend
-    if backend == "auto" and not download_only:
+    if backend == "auto":
         from agent_cli.server.whisper.backends import detect_backend  # noqa: PLC0415
 
         resolved_backend = detect_backend()
@@ -370,13 +377,26 @@ def whisper_cmd(  # noqa: PLR0912, PLR0915
         for model_name in model:
             console.print(f"  Downloading [cyan]{model_name}[/cyan]...")
             try:
-                from faster_whisper import WhisperModel  # noqa: PLC0415
-
-                _ = WhisperModel(
-                    model_name,
-                    device="cpu",  # Don't need GPU for download
-                    download_root=str(cache_dir) if cache_dir else None,
-                )
+                if resolved_backend == "transformers":
+                    from agent_cli.server.whisper.backends.transformers import (  # noqa: PLC0415
+                        download_model as download_transformers_model,
+                    )
+
+                    download_transformers_model(model_name, cache_dir=cache_dir)
+                elif resolved_backend == "mlx":
+                    from agent_cli.server.whisper.backends.mlx import (  # noqa: PLC0415
+                        download_model as download_mlx_model,
+                    )
+
+                    download_mlx_model(model_name)
+                else:
+                    from faster_whisper import WhisperModel  # noqa: PLC0415
+
+                    _ = WhisperModel(
+                        model_name,
+                        device="cpu",  # Don't need GPU for download
+                        download_root=str(cache_dir) if cache_dir else None,
+                    )
                 console.print(f"  [green]✓[/green] Downloaded {model_name}")
             except Exception as e:
                 err_console.print(f"  [red]✗[/red] Failed to download {model_name}: {e}")

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/server/whisper/backends/__init__.py

@@ -17,7 +17,7 @@ from agent_cli.server.whisper.backends.base import (
 
 logger = logging.getLogger(__name__)
 
-BackendType = Literal["faster-whisper", "mlx", "auto"]
+BackendType = Literal["faster-whisper", "mlx", "transformers", "auto"]
 
 
 def detect_backend() -> Literal["faster-whisper", "mlx"]:
@@ -76,6 +76,13 @@ def create_backend(
 
         return FasterWhisperBackend(config)
 
+    if backend_type == "transformers":
+        from agent_cli.server.whisper.backends.transformers import (  # noqa: PLC0415
+            TransformersWhisperBackend,
+        )
+
+        return TransformersWhisperBackend(config)
+
     msg = f"Unknown backend type: {backend_type}"
     raise ValueError(msg)
 

{agent_cli-0.73.0 → agent_cli-0.75.0}/agent_cli/server/whisper/backends/mlx.py

@@ -55,6 +55,16 @@ def _resolve_mlx_model_name(model_name: str) -> str:
     return model_name
 
 
+def download_model(model_name: str) -> str:
+    """Download an MLX Whisper model and return the resolved repo name."""
+    import mlx.core as mx  # noqa: PLC0415
+    from mlx_whisper.transcribe import ModelHolder  # noqa: PLC0415
+
+    resolved_model = _resolve_mlx_model_name(model_name)
+    ModelHolder.get_model(resolved_model, mx.float16)
+    return resolved_model
+
+
 def _pcm_to_float(audio_bytes: bytes) -> NDArray[np.float32]:
     """Convert 16-bit PCM audio bytes to float32 array normalized to [-1, 1]."""
     import numpy as np  # noqa: PLC0415