agent-cli 0.72.7__tar.gz → 0.74.0__tar.gz

{agent_cli-0.72.7 → agent_cli-0.74.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-cli
-Version: 0.72.7
+Version: 0.74.0
 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'
@@ -2066,8 +2064,12 @@ uv tool install "agent-cli[vad]" -p 3.13
 │                                [env var: OPENAI_API_KEY]                               │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ LLM Configuration ────────────────────────────────────────────────────────────────────╮
-│ --embedding-model        TEXT  Embedding model to use for vectorization.               │
-│                                [default: text-embedding-3-small]                       │
+│ --embedding-base-url        TEXT  Base URL for embedding API. Falls back to            │
+│                                   --openai-base-url if not set. Useful when using      │
+│                                   different providers for chat vs embeddings.          │
+│                                   [env var: EMBEDDING_BASE_URL]                        │
+│ --embedding-model           TEXT  Embedding model to use for vectorization.            │
+│                                   [default: text-embedding-3-small]                    │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ Server Configuration ─────────────────────────────────────────────────────────────────╮
 │ --host        TEXT     Host/IP to bind API servers to.                                 │
@@ -2229,8 +2231,12 @@ The `memory proxy` command is the core feature—a middleware server that gives
 │                                [env var: OPENAI_API_KEY]                               │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ LLM Configuration ────────────────────────────────────────────────────────────────────╮
-│ --embedding-model        TEXT  Embedding model to use for vectorization.               │
-│                                [default: text-embedding-3-small]                       │
+│ --embedding-base-url        TEXT  Base URL for embedding API. Falls back to            │
+│                                   --openai-base-url if not set. Useful when using      │
+│                                   different providers for chat vs embeddings.          │
+│                                   [env var: EMBEDDING_BASE_URL]                        │
+│ --embedding-model           TEXT  Embedding model to use for vectorization.            │
+│                                   [default: text-embedding-3-small]                    │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ Server Configuration ─────────────────────────────────────────────────────────────────╮
 │ --host        TEXT     Host/IP to bind API servers to.                                 │

{agent_cli-0.72.7 → agent_cli-0.74.0}/README.md

@@ -1974,8 +1974,12 @@ uv tool install "agent-cli[vad]" -p 3.13
 │                                [env var: OPENAI_API_KEY]                               │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ LLM Configuration ────────────────────────────────────────────────────────────────────╮
-│ --embedding-model        TEXT  Embedding model to use for vectorization.               │
-│                                [default: text-embedding-3-small]                       │
+│ --embedding-base-url        TEXT  Base URL for embedding API. Falls back to            │
+│                                   --openai-base-url if not set. Useful when using      │
+│                                   different providers for chat vs embeddings.          │
+│                                   [env var: EMBEDDING_BASE_URL]                        │
+│ --embedding-model           TEXT  Embedding model to use for vectorization.            │
+│                                   [default: text-embedding-3-small]                    │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ Server Configuration ─────────────────────────────────────────────────────────────────╮
 │ --host        TEXT     Host/IP to bind API servers to.                                 │
@@ -2137,8 +2141,12 @@ The `memory proxy` command is the core feature—a middleware server that gives
 │                                [env var: OPENAI_API_KEY]                               │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ LLM Configuration ────────────────────────────────────────────────────────────────────╮
-│ --embedding-model        TEXT  Embedding model to use for vectorization.               │
-│                                [default: text-embedding-3-small]                       │
+│ --embedding-base-url        TEXT  Base URL for embedding API. Falls back to            │
+│                                   --openai-base-url if not set. Useful when using      │
+│                                   different providers for chat vs embeddings.          │
+│                                   [env var: EMBEDDING_BASE_URL]                        │
+│ --embedding-model           TEXT  Embedding model to use for vectorization.            │
+│                                   [default: text-embedding-3-small]                    │
 ╰────────────────────────────────────────────────────────────────────────────────────────╯
 ╭─ Server Configuration ─────────────────────────────────────────────────────────────────╮
 │ --host        TEXT     Host/IP to bind API servers to.                                 │

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/_extras.json

@@ -19,12 +19,6 @@
       "pydantic_ai"
     ]
   ],
-  "llm-core": [
-    "Minimal LLM support for Docker images",
-    [
-      "pydantic_ai"
-    ]
-  ],
   "memory": [
     "Long-term memory proxy",
     [

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/agents/memory/proxy.py

@@ -23,6 +23,7 @@ def proxy(
         rich_help_panel="Memory Configuration",
     ),
     openai_base_url: str | None = opts.OPENAI_BASE_URL,
+    embedding_base_url: str | None = opts.EMBEDDING_BASE_URL,
     embedding_model: str = opts.EMBEDDING_MODEL,
     openai_api_key: str | None = opts.OPENAI_API_KEY,
     default_top_k: int = typer.Option(
@@ -135,12 +136,15 @@ def proxy(
     entries_dir, _ = ensure_store_dirs(memory_path)
     if openai_base_url is None:
         openai_base_url = constants.DEFAULT_OPENAI_BASE_URL
+    effective_embedding_url = embedding_base_url or openai_base_url
 
     console.print(f"[bold green]Starting Memory Proxy on {host}:{port}[/bold green]")
     console.print(f"  💾 Memory store: [blue]{memory_path}[/blue]")
     console.print(f"  📁 Entries: [blue]{entries_dir}[/blue]")
     console.print(f"  🤖 Backend: [blue]{openai_base_url}[/blue]")
-    console.print(f"  🧠 Embeddings: Using [blue]{embedding_model}[/blue]")
+    console.print(
+        f"  🧠 Embeddings: [blue]{embedding_model}[/blue] via [blue]{effective_embedding_url}[/blue]",
+    )
     console.print(f"  🔍 Memory top_k: [blue]{default_top_k}[/blue] entries per query")
     console.print(f"  🧹 Max entries per conversation: [blue]{max_entries}[/blue]")
     console.print(
@@ -154,6 +158,7 @@ def proxy(
     fastapi_app = create_app(
         memory_path,
         openai_base_url,
+        embedding_base_url=effective_embedding_url,
         embedding_model=embedding_model,
         embedding_api_key=openai_api_key,
         chat_api_key=openai_api_key,

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/agents/rag_proxy.py

@@ -32,6 +32,7 @@ def rag_proxy(
         rich_help_panel="RAG Configuration",
     ),
     openai_base_url: str | None = opts.OPENAI_BASE_URL,
+    embedding_base_url: str | None = opts.EMBEDDING_BASE_URL,
     embedding_model: str = opts.EMBEDDING_MODEL,
     openai_api_key: str | None = opts.OPENAI_API_KEY,
     limit: int = typer.Option(
@@ -138,22 +139,26 @@ def rag_proxy(
 
     if openai_base_url is None:
         openai_base_url = constants.DEFAULT_OPENAI_BASE_URL
+    effective_embedding_url = embedding_base_url or openai_base_url
 
     console.print(f"[bold green]Starting RAG Proxy on {host}:{port}[/bold green]")
     console.print(f"  📂 Docs: [blue]{docs_folder}[/blue]")
     console.print(f"  💾 DB: [blue]{chroma_path}[/blue]")
     console.print(f"  🤖 Backend: [blue]{openai_base_url}[/blue]")
-    console.print(f"  🧠 Embeddings: Using [blue]{embedding_model}[/blue]")
+    console.print(
+        f"  🧠 Embeddings: [blue]{embedding_model}[/blue] via [blue]{effective_embedding_url}[/blue]",
+    )
     console.print(f"  🔍 Limit: [blue]{limit}[/blue] chunks per query")
 
     fastapi_app = create_app(
         docs_folder,
         chroma_path,
         openai_base_url,
-        embedding_model,
-        openai_api_key,
-        openai_api_key,
-        limit,
+        embedding_base_url=effective_embedding_url,
+        embedding_model=embedding_model,
+        embedding_api_key=openai_api_key,
+        chat_api_key=openai_api_key,
+        limit=limit,
         enable_rag_tools=enable_rag_tools,
     )
 

{agent_cli-0.72.7 → agent_cli-0.74.0}/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")

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/memory/api.py

@@ -22,6 +22,7 @@ LOGGER = logging.getLogger(__name__)
 def create_app(
     memory_path: Path,
     openai_base_url: str,
+    embedding_base_url: str | None = None,
     embedding_model: str = DEFAULT_OPENAI_EMBEDDING_MODEL,
     embedding_api_key: str | None = None,
     chat_api_key: str | None = None,
@@ -39,6 +40,7 @@ def create_app(
     client = MemoryClient(
         memory_path=memory_path,
         openai_base_url=openai_base_url,
+        embedding_base_url=embedding_base_url,
         embedding_model=embedding_model,
         embedding_api_key=embedding_api_key,
         chat_api_key=chat_api_key,

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/memory/client.py

@@ -41,6 +41,7 @@ class MemoryClient:
         self,
         memory_path: Path,
         openai_base_url: str,
+        embedding_base_url: str | None = None,
         embedding_model: str = DEFAULT_OPENAI_EMBEDDING_MODEL,
         embedding_api_key: str | None = None,
         chat_api_key: str | None = None,
@@ -74,7 +75,7 @@ class MemoryClient:
         self.collection: Collection = init_memory_collection(
             self.memory_path,
             embedding_model=embedding_model,
-            openai_base_url=self.openai_base_url,
+            openai_base_url=embedding_base_url or self.openai_base_url,
             openai_api_key=embedding_api_key,
         )
 

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/opts.py

@@ -109,6 +109,14 @@ EMBEDDING_MODEL: str = typer.Option(
     help="Embedding model to use for vectorization.",
     rich_help_panel="LLM Configuration",
 )
+EMBEDDING_BASE_URL: str | None = typer.Option(
+    None,
+    "--embedding-base-url",
+    envvar="EMBEDDING_BASE_URL",
+    help="Base URL for embedding API. Falls back to `--openai-base-url` if not set. "
+    "Useful when using different providers for chat vs embeddings.",
+    rich_help_panel="LLM Configuration",
+)
 
 # --- ASR (Audio) Configuration ---
 # General ASR

{agent_cli-0.72.7 → agent_cli-0.74.0}/agent_cli/rag/api.py

@@ -32,6 +32,7 @@ def create_app(
     docs_folder: Path,
     chroma_path: Path,
     openai_base_url: str,
+    embedding_base_url: str | None = None,
     embedding_model: str = DEFAULT_OPENAI_EMBEDDING_MODEL,
     embedding_api_key: str | None = None,
     chat_api_key: str | None = None,
@@ -47,7 +48,7 @@ def create_app(
         chroma_path,
         name="docs",
         embedding_model=embedding_model,
-        openai_base_url=openai_base_url,
+        openai_base_url=embedding_base_url or openai_base_url,
         openai_api_key=embedding_api_key,
     )
 

{agent_cli-0.72.7 → agent_cli-0.74.0}/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"]),

{agent_cli-0.72.7 → agent_cli-0.74.0}/docker/docker-compose.yml

@@ -46,6 +46,7 @@
 #   RAG_LIMIT          - Number of document chunks per query (default: 3)
 #   RAG_ENABLE_TOOLS   - Enable read_full_document tool (default: true)
 #   EMBEDDING_MODEL    - Embedding model for RAG/memory (default: text-embedding-3-small)
+#   EMBEDDING_BASE_URL - Separate endpoint for embeddings (falls back to OPENAI_BASE_URL)
 #
 #   MEMORY_PORT        - Memory proxy port (default: 8100)
 #   MEMORY_TOP_K       - Number of memories per query (default: 5)
@@ -262,6 +263,7 @@ services:
       - RAG_LIMIT=${RAG_LIMIT:-3}
       - RAG_ENABLE_TOOLS=${RAG_ENABLE_TOOLS:-true}
       - EMBEDDING_MODEL=${EMBEDDING_MODEL:-text-embedding-3-small}
+      - EMBEDDING_BASE_URL=${EMBEDDING_BASE_URL:-}
       - LOG_LEVEL=${LOG_LEVEL:-info}
       # OpenAI-compatible backend configuration
       - OPENAI_BASE_URL=${OPENAI_BASE_URL:-http://ollama:11434/v1}
@@ -295,6 +297,7 @@ services:
       - MEMORY_SUMMARIZATION=${MEMORY_SUMMARIZATION:-true}
       - MEMORY_GIT_VERSIONING=${MEMORY_GIT_VERSIONING:-true}
       - EMBEDDING_MODEL=${EMBEDDING_MODEL:-text-embedding-3-small}
+      - EMBEDDING_BASE_URL=${EMBEDDING_BASE_URL:-}
       - LOG_LEVEL=${LOG_LEVEL:-info}
       # OpenAI-compatible backend configuration
       - OPENAI_BASE_URL=${OPENAI_BASE_URL:-http://ollama:11434/v1}

{agent_cli-0.72.7 → agent_cli-0.74.0}/docker/transcribe-proxy.Dockerfile

@@ -36,7 +36,7 @@ COPY pyproject.toml uv.lock README.md ./
 COPY .git ./.git
 COPY agent_cli ./agent_cli
 COPY scripts ./scripts
-RUN uv sync --frozen --no-dev --no-editable --extra server --extra wyoming --extra llm-core
+RUN uv sync --frozen --no-dev --no-editable --extra server --extra wyoming --extra llm
 
 # =============================================================================
 # Runtime stage - minimal Alpine image

{agent_cli-0.72.7 → agent_cli-0.74.0}/docs/commands/install-extras.md

@@ -35,7 +35,6 @@ Available extras:
 | `faster-whisper` | TODO: add description |
 | `kokoro` | TODO: add description |
 | `llm` | LLM framework (pydantic-ai) |
-| `llm-core` | Minimal LLM support for Docker images |
 | `memory` | Long-term memory proxy |
 | `mlx-whisper` | TODO: add description |
 | `piper` | TODO: add description |

{agent_cli-0.72.7 → agent_cli-0.74.0}/docs/commands/memory.md

@@ -98,6 +98,7 @@ agent-cli chat --openai-base-url http://localhost:8100/v1 --llm-provider openai
 
 | Option | Default | Description |
 |--------|---------|-------------|
+| `--embedding-base-url` | - | Base URL for embedding API. Falls back to `--openai-base-url` if not set. Useful when using different providers for chat vs embeddings. |
 | `--embedding-model` | `text-embedding-3-small` | Embedding model to use for vectorization. |
 
 ### Server Configuration

{agent_cli-0.72.7 → agent_cli-0.74.0}/docs/commands/rag-proxy.md

@@ -78,6 +78,7 @@ agent-cli chat --openai-base-url http://localhost:8000/v1 --llm-provider openai
 
 | Option | Default | Description |
 |--------|---------|-------------|
+| `--embedding-base-url` | - | Base URL for embedding API. Falls back to `--openai-base-url` if not set. Useful when using different providers for chat vs embeddings. |
 | `--embedding-model` | `text-embedding-3-small` | Embedding model to use for vectorization. |
 
 ### Server Configuration

{agent_cli-0.72.7 → agent_cli-0.74.0}/pyproject.toml

@@ -33,15 +33,14 @@ Homepage = "https://github.com/basnijholt/agent-cli"
 # Provider extras (previously base dependencies, now optional)
 wyoming = ["wyoming>=1.5.2"]  # Wyoming protocol only (for servers)
 audio = ["numpy", "sounddevice>=0.4.6", "agent-cli[wyoming]"]  # Local audio recording
-llm-core = ["pydantic-ai-slim[openai,google]>=0.1.1"]  # Minimal LLM support for Docker
-llm = ["pydantic-ai-slim[openai,google,duckduckgo,vertexai]>=0.1.1"]  # Full LLM with tools
+llm = ["pydantic-ai-slim[openai,google,duckduckgo,vertexai]>=0.1.1"]  # LLM framework with tools
 
 # Feature extras
 server = [
     "fastapi[standard]",
 ]
 rag = [
-    "agent-cli[llm-core]",  # pydantic-ai for LLM integration
+    "pydantic-ai-slim[openai,google]>=0.1.1",  # pydantic-ai for LLM integration
     "fastapi[standard]",
     "chromadb>=0.4.22",
     "openai>=1.0.0",  # Required by chromadb's OpenAIEmbeddingFunction
@@ -49,11 +48,11 @@ rag = [
     "huggingface-hub>=0.20.0",
     "transformers>=4.30.0",
     "watchfiles>=0.21.0",
-    # Until here same as 'memory' extras (except llm-core)
+    # Until here same as 'memory' extras
     "markitdown[docx,pdf,pptx]>=0.1.3",
 ]
 memory = [
-    "agent-cli[llm-core]",  # pydantic-ai for LLM integration
+    "pydantic-ai-slim[openai,google]>=0.1.1",  # pydantic-ai for LLM integration
     "fastapi[standard]",
     "chromadb>=0.4.22",
     "openai>=1.0.0",  # Required by chromadb's OpenAIEmbeddingFunction
@@ -61,7 +60,7 @@ memory = [
     "huggingface-hub>=0.20.0",
     "transformers>=4.30.0",
     "watchfiles>=0.21.0",
-    # Until here same as 'rag' extras (except llm-core)
+    # Until here same as 'rag' extras
     "pyyaml>=6.0.0",
 ]
 vad = [

{agent_cli-0.72.7 → agent_cli-0.74.0}/uv.lock

@@ -71,9 +71,6 @@ kokoro = [
 llm = [
     { name = "pydantic-ai-slim", extra = ["duckduckgo", "google", "openai", "vertexai"] },
 ]
-llm-core = [
-    { name = "pydantic-ai-slim", extra = ["google", "openai"] },
-]
 memory = [
     { name = "chromadb" },
     { name = "fastapi", extra = ["standard"] },
@@ -178,7 +175,6 @@ requires-dist = [
     { name = "psutil", marker = "sys_platform == 'win32'" },
     { name = "pydantic" },
     { name = "pydantic-ai-slim", extras = ["duckduckgo", "google", "openai", "vertexai"], marker = "extra == 'llm'", specifier = ">=0.1.1" },
-    { name = "pydantic-ai-slim", extras = ["google", "openai"], marker = "extra == 'llm-core'", specifier = ">=0.1.1" },
     { name = "pydantic-ai-slim", extras = ["google", "openai"], marker = "extra == 'memory'", specifier = ">=0.1.1" },
     { name = "pydantic-ai-slim", extras = ["google", "openai"], marker = "extra == 'rag'", specifier = ">=0.1.1" },
     { name = "pylint", marker = "extra == 'dev'", specifier = ">=3.0.0" },
@@ -212,7 +208,7 @@ requires-dist = [
     { name = "wyoming", marker = "extra == 'wyoming'", specifier = ">=1.5.2" },
     { name = "zensical", marker = "extra == 'dev'" },
 ]
-provides-extras = ["audio", "dev", "faster-whisper", "kokoro", "llm", "llm-core", "memory", "mlx-whisper", "piper", "rag", "server", "speed", "test", "vad", "wyoming"]
+provides-extras = ["audio", "dev", "faster-whisper", "kokoro", "llm", "memory", "mlx-whisper", "piper", "rag", "server", "speed", "test", "vad", "wyoming"]
 
 [package.metadata.requires-dev]
 dev = [