alloy-runtime-cli 0.2.63__tar.gz → 0.2.65__tar.gz

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alloy-runtime-cli
-Version: 0.2.63
+Version: 0.2.65
 Summary: Alloy Runtime CLI - Command-line interface for Alloy Runtime server
 Keywords: alloy,cli,python,terminal,workflow
 Classifier: Development Status :: 3 - Alpha
@@ -16,8 +16,8 @@ Requires-Dist: textual==8.2.3
 Requires-Dist: genson==1.3.0
 Requires-Dist: pyperclip==1.11.0
 Requires-Dist: prompt-toolkit==3.0.52
-Requires-Dist: alloy-runtime-sdk==0.2.63
-Requires-Dist: alloy-runtime-types==0.2.63
+Requires-Dist: alloy-runtime-sdk==0.2.65
+Requires-Dist: alloy-runtime-types==0.2.65
 Requires-Dist: trogon==0.6.0
 
 # alloy-runtime-cli
@@ -65,6 +65,10 @@ Use `alloy --format compact` for the existing psql-style human-readable record v
 Use `alloy --format table` for Rich table output in the non-TUI CLI.
 Use `alloy --format json` for automation-friendly machine-readable output.
 
+Running `alloy` without a subcommand launches the Textual TUI. TUI chat uses the
+standard non-streaming generation path, so it works with agents that have tools
+configured. Responses appear once the full turn completes rather than token-by-token.
+
 For live model-execution inspection, use:
 
 ```bash
@@ -101,4 +105,3 @@ Use `alloy login` to save an API key locally, or set the environment variables w
 running on a VPS.
 
 CLI release trigger: no functional changes.
-

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/README.md

@@ -43,6 +43,10 @@ Use `alloy --format compact` for the existing psql-style human-readable record v
 Use `alloy --format table` for Rich table output in the non-TUI CLI.
 Use `alloy --format json` for automation-friendly machine-readable output.
 
+Running `alloy` without a subcommand launches the Textual TUI. TUI chat uses the
+standard non-streaming generation path, so it works with agents that have tools
+configured. Responses appear once the full turn completes rather than token-by-token.
+
 For live model-execution inspection, use:
 
 ```bash
@@ -79,4 +83,3 @@ Use `alloy login` to save an API key locally, or set the environment variables w
 running on a VPS.
 
 CLI release trigger: no functional changes.
-

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/alloy_runtime_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alloy-runtime-cli
-Version: 0.2.63
+Version: 0.2.65
 Summary: Alloy Runtime CLI - Command-line interface for Alloy Runtime server
 Keywords: alloy,cli,python,terminal,workflow
 Classifier: Development Status :: 3 - Alpha
@@ -16,8 +16,8 @@ Requires-Dist: textual==8.2.3
 Requires-Dist: genson==1.3.0
 Requires-Dist: pyperclip==1.11.0
 Requires-Dist: prompt-toolkit==3.0.52
-Requires-Dist: alloy-runtime-sdk==0.2.63
-Requires-Dist: alloy-runtime-types==0.2.63
+Requires-Dist: alloy-runtime-sdk==0.2.65
+Requires-Dist: alloy-runtime-types==0.2.65
 Requires-Dist: trogon==0.6.0
 
 # alloy-runtime-cli
@@ -65,6 +65,10 @@ Use `alloy --format compact` for the existing psql-style human-readable record v
 Use `alloy --format table` for Rich table output in the non-TUI CLI.
 Use `alloy --format json` for automation-friendly machine-readable output.
 
+Running `alloy` without a subcommand launches the Textual TUI. TUI chat uses the
+standard non-streaming generation path, so it works with agents that have tools
+configured. Responses appear once the full turn completes rather than token-by-token.
+
 For live model-execution inspection, use:
 
 ```bash
@@ -101,4 +105,3 @@ Use `alloy login` to save an API key locally, or set the environment variables w
 running on a VPS.
 
 CLI release trigger: no functional changes.
-

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/alloy_runtime_cli.egg-info/requires.txt

@@ -4,6 +4,6 @@ textual==8.2.3
 genson==1.3.0
 pyperclip==1.11.0
 prompt-toolkit==3.0.52
-alloy-runtime-sdk==0.2.63
-alloy-runtime-types==0.2.63
+alloy-runtime-sdk==0.2.65
+alloy-runtime-types==0.2.65
 trogon==0.6.0

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/commands/knowledge/documents/list/command.py

@@ -19,6 +19,11 @@ def knowledge_documents_list_command(
         "--status",
         help="Filter by status: queued|processing|completed|failed|duplicate|filtered",
     ),
+    external_id: str | None = typer.Option(
+        None,
+        "--external-id",
+        help="Filter by exact external_id match",
+    ),
     limit: int = typer.Option(
         50,
         "-l",
@@ -40,6 +45,7 @@ def knowledge_documents_list_command(
         alloy knowledge documents list
         alloy knowledge documents list -C my-collection
         alloy knowledge documents list -s completed -l 20
+        alloy knowledge documents list --external-id crm-123
         alloy knowledge documents list -s duplicate
         alloy knowledge documents list -s filtered
         alloy knowledge documents list --status failed
@@ -47,6 +53,7 @@ def knowledge_documents_list_command(
     _execute_list(
         collection_id=collection,
         processing_status=status,
+        external_id=external_id,
         limit=limit,
         offset=offset,
     )
@@ -56,6 +63,7 @@ def knowledge_documents_list_command(
 async def _execute_list(
     collection_id: str | None,
     processing_status: str | None,
+    external_id: str | None,
     limit: int,
     offset: int,
 ) -> None:
@@ -66,6 +74,7 @@ async def _execute_list(
             offset=offset,
             collection_id=collection_id,
             processing_status=processing_status,
+            external_id=external_id,
         )
 
     present_documents_list(response)

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/commands/knowledge/search/command.py

@@ -1,11 +1,14 @@
 """Knowledge search command implementation."""
 
-import json
-from typing import Any, Optional
+from typing import Optional
 from uuid import UUID
 
 import typer
 
+from alloy_runtime_types.parsing.filter_parser import (
+    FilterParseError,
+    parse_filter_spec,
+)
 from cli.commands.flag_utils import validate_uuid
 from cli.commands.knowledge.search.presenter import present_search_results
 from cli.infrastructure.command import async_command, authenticated_client
@@ -166,17 +169,14 @@ def _execute_search_from_flags(
             decomposition_template, "decomposition-template"
         )
 
-    parsed_metadata_filter: dict[str, Any] | None = None
+    parsed_metadata_filter = None
     if metadata_filter:
         try:
-            parsed_metadata_filter = json.loads(metadata_filter)
-            if not isinstance(parsed_metadata_filter, dict):
-                raise typer.BadParameter(
-                    "--metadata-filter must be a JSON object, not "
-                    f"{type(parsed_metadata_filter).__name__}"
-                )
-        except json.JSONDecodeError as e:
-            raise typer.BadParameter(f"Invalid JSON in --metadata-filter: {e}")
+            parsed_metadata_filter = parse_filter_spec(metadata_filter)
+        except FilterParseError as error:
+            raise typer.BadParameter(
+                f"Invalid JSON in --metadata-filter: {error}"
+            ) from error
 
     request = SearchRequest(
         query=query,

alloy_runtime_cli-0.2.65/cli/commands/knowledge/synthesis/list/presenter.py

@@ -0,0 +1,87 @@
+"""Presenter for synthesis list output."""
+
+from datetime import datetime
+from typing import Any, cast
+
+from pydantic import BaseModel
+from alloy_runtime_types.dtos.knowledge import ListSynthesesResponse, SynthesisSummary
+from cli.infrastructure.formatting.fields import format_datetime, format_uuid
+from cli.infrastructure.output import OutputService
+from cli.infrastructure.renderers.list_renderer import ColumnConfig
+
+
+def _format_stale(value: bool) -> str:
+    return "stale" if value else "current"
+
+
+def _format_collections(value: list[Any]) -> Any:
+    if not value:
+        return cast(str, None)
+    return ", ".join(str(collection_id) for collection_id in value)
+
+
+def _latest_heartbeat_at(synthesis: SynthesisSummary) -> datetime | None:
+    queue_metadata = synthesis.queue_metadata
+    if queue_metadata is None:
+        return None
+    if isinstance(queue_metadata, dict):
+        queue_metadata_dict = cast(dict[str, Any], queue_metadata)
+        latest_heartbeat = queue_metadata_dict.get("last_heartbeat_at")
+        if isinstance(latest_heartbeat, datetime):
+            return latest_heartbeat
+        return None
+    return queue_metadata.last_heartbeat_at
+
+
+def _format_heartbeat(synthesis: SynthesisSummary) -> str:
+    if synthesis.status in {"accepted", "queued"}:
+        return "pending"
+    if synthesis.status in {"worker_assigned", "processing"}:
+        return "active" if _latest_heartbeat_at(synthesis) is not None else "missing"
+    return "stopped"
+
+
+class SynthesisListRow(BaseModel):
+    id: Any
+    title: str
+    status: str
+    heartbeat: str
+    latest_heartbeat_at: datetime | None
+    is_stale: bool
+    source_chunk_count: int
+    collection_ids: list[Any]
+    created_at: datetime
+
+
+def present_syntheses_list(response: ListSynthesesResponse) -> None:
+    rows = [
+        SynthesisListRow(
+            id=synthesis.id,
+            title=synthesis.title,
+            status=synthesis.status,
+            heartbeat=_format_heartbeat(synthesis),
+            latest_heartbeat_at=_latest_heartbeat_at(synthesis),
+            is_stale=synthesis.is_stale,
+            source_chunk_count=synthesis.source_chunk_count,
+            collection_ids=synthesis.collection_ids,
+            created_at=synthesis.created_at,
+        )
+        for synthesis in response.syntheses
+    ]
+    items = cast(list[BaseModel], rows)
+    OutputService.get().table(
+        items,
+        f"Syntheses ({response.total})",
+        [
+            ColumnConfig("id", "ID", lambda value: format_uuid(value, short=True)),
+            ColumnConfig("title", "Title"),
+            ColumnConfig("status", "Status"),
+            ColumnConfig("heartbeat", "Heartbeat"),
+            ColumnConfig("latest_heartbeat_at", "Latest Heartbeat", format_datetime),
+            ColumnConfig("is_stale", "Stale", _format_stale),
+            ColumnConfig("source_chunk_count", "Sources"),
+            ColumnConfig("collection_ids", "Collections", _format_collections),
+            ColumnConfig("created_at", "Created", format_datetime),
+        ],
+        raw_payload=response,
+    )

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/infrastructure/filter_parser.py

@@ -186,10 +186,6 @@ def parse_single_filter(expr: str) -> ConditionSpec:
         else:  # auto
             value = _parse_auto_value(raw_value)
 
-    # For exists/not_exists, value should be None (or True for consistency)
-    if op_name in ("exists", "not_exists"):
-        value = True
-
     # Cast op_name to the literal type since we've validated it above
     op_literal = cast(ConditionOperator, op_name)
     return ConditionSpec(field=field, op=op_literal, value=value)

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/tui/chat/commands/regenerate.py

@@ -1,7 +1,6 @@
 """Command for regenerating the last AI response."""
 
 from dataclasses import dataclass
-from typing import Callable
 from uuid import UUID
 
 from alloy_runtime_sdk.api_client.client import ApiClient
@@ -28,17 +27,16 @@ class RegenerateCommand:
     - Model mode: Uses provider_key/provider_model_name/system_instruction directly
 
     This command:
-    1. Transitions store to STREAMING phase
+    1. Transitions store to SENDING phase
     2. Sends a regenerate request to the API
-    3. Streams the new response chunks via callbacks (content and thinking)
-    4. Transitions back to IDLE when complete
+    3. Waits for the final non-streaming response
+    4. Leaves the store in SENDING on success so the caller can reload persisted
+       messages before clearing the waiting state
 
     Attributes:
         client: Server client for API calls.
         store: Chat store for state management.
         session_id: ID of the current chat session.
-        on_chunk: Callback invoked with each streamed text chunk.
-        on_thinking_chunk: Callback invoked with each reasoning/thinking chunk.
         agent_id: ID of the agent to use for regeneration (agent mode).
         provider_key: Provider key (model mode).
         provider_model_name: Model name (model mode).
@@ -48,8 +46,6 @@ class RegenerateCommand:
     client: ApiClient
     store: ChatStore
     session_id: UUID
-    on_chunk: Callable[[str], None]
-    on_thinking_chunk: Callable[[str], None] | None = None
     # Agent mode
     agent_id: UUID | None = None
     # Model mode
@@ -64,9 +60,8 @@ class RegenerateCommand:
             CommandResult indicating success or failure.
         """
         logger.info("regenerate_starting", session_id=str(self.session_id))
-        self.store.start_streaming()
+        self.store.start_generation()
 
-        was_cancelled = False
         try:
             # Build model source based on mode
             if self.agent_id is not None:
@@ -77,6 +72,7 @@ class RegenerateCommand:
                     provider_model_name=self.provider_model_name,
                 )
             else:
+                self.store.finish_generation()
                 return CommandResult[None](
                     success=False,
                     error="No agent or model configured for this session",
@@ -86,7 +82,7 @@ class RegenerateCommand:
                 model_source=model_source,
                 chat_session_id=self.session_id,
                 regenerate=True,
-                stream=True,
+                stream=False,
                 tags=[],
                 # Only include system instruction for model mode
                 system_instruction=self.system_instruction
@@ -94,27 +90,12 @@ class RegenerateCommand:
                 else None,
             )
 
-            async for chunk in self.client.generate_text_stream(request):
-                # Check for cancellation request
-                if self.store.state.cancel_requested:
-                    logger.info("regenerate_cancelled_by_user")
-                    was_cancelled = True
-                    return CommandResult[None](success=True, data=None)
-
-                # Handle thinking/reasoning chunks
-                if chunk.chunk_type == "thinking" and chunk.reasoning_content:
-                    if self.on_thinking_chunk:
-                        self.on_thinking_chunk(chunk.reasoning_content)
-                # Handle regular text content chunks
-                elif chunk.content:
-                    self.on_chunk(chunk.content)
+            await self.client.generate_text(request)
 
             logger.info("regenerate_completed")
             return CommandResult[None](success=True, data=None)
 
         except Exception as e:
             logger.warning("regenerate_failed", error=str(e))
+            self.store.finish_generation()
             return CommandResult[None](success=False, error=str(e))
-
-        finally:
-            self.store.finish_streaming(cancelled=was_cancelled)

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/tui/chat/commands/reload_session.py

@@ -59,5 +59,4 @@ class ReloadSessionCommand:
             )
 
         except Exception as e:
-            # Silently fail for reload - the user already saw the response
             return CommandResult[ReloadSessionResult](success=False, error=str(e))

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/tui/chat/commands/send_message.py

@@ -1,4 +1,4 @@
-"""Command for sending a message and streaming the response."""
+"""Command for sending a message and waiting for the final response."""
 
 from dataclasses import dataclass
 from typing import Callable
@@ -24,7 +24,7 @@ logger = get_logger(__name__)
 
 @dataclass
 class SendMessageCommand:
-    """Command to send a user message and stream the AI response.
+    """Command to send a user message and await the final AI response.
 
     Supports two modes:
     - Agent mode: Uses agent_id to determine model and system instruction
@@ -38,18 +38,17 @@ class SendMessageCommand:
 
     This command:
     1. Resolves any injection patterns in the message
-    2. Transitions store to STREAMING phase
+    2. Transitions store to SENDING phase
     3. Builds and sends the generation request
-    4. Streams response chunks via callbacks (content and thinking)
-    5. Transitions back to IDLE when complete
+    4. Waits for the final response from the non-streaming API
+    5. Leaves the store in SENDING on success so the caller can reload persisted
+       messages before clearing the optimistic pending state
 
     Attributes:
         client: Server client for API calls.
         store: Chat store for state management.
         session_id: ID of the current chat session.
         user_message: The user's message to send.
-        on_chunk: Callback invoked with each streamed text chunk.
-        on_thinking_chunk: Callback invoked with each reasoning/thinking chunk.
         agent_id: ID of the agent to use for generation (agent mode).
         provider_key: Provider key (model mode).
         provider_model_name: Model name (model mode).
@@ -62,8 +61,6 @@ class SendMessageCommand:
     store: ChatStore
     session_id: UUID
     user_message: str
-    on_chunk: Callable[[str], None]
-    on_thinking_chunk: Callable[[str], None] | None = None
     # Agent mode
     agent_id: UUID | None = None
     # Model mode
@@ -75,7 +72,7 @@ class SendMessageCommand:
     on_injection_resolved: Callable[[ResolvedMessage], None] | None = None
 
     async def execute(self) -> CommandResult[None]:
-        """Execute the send message and stream operation.
+        """Execute the send message operation.
 
         Returns:
             CommandResult indicating success or failure.
@@ -123,11 +120,9 @@ class SendMessageCommand:
                     error=f"Failed to resolve injections: {e}",
                 )
 
-        # Start streaming with the user message shown optimistically
+        # Start generation with the user message shown optimistically
         # Note: We show the original message, not the resolved one
-        self.store.start_streaming(user_message=self.user_message)
-
-        was_cancelled = False
+        self.store.start_generation(user_message=self.user_message)
         try:
             # Build model source based on mode
             if self.agent_id is not None:
@@ -138,6 +133,7 @@ class SendMessageCommand:
                     provider_model_name=self.provider_model_name,
                 )
             else:
+                self.store.finish_generation()
                 return CommandResult[None](
                     success=False,
                     error="No agent or model configured for this session",
@@ -146,7 +142,7 @@ class SendMessageCommand:
             request = GenerateTextRequest(
                 model_source=model_source,
                 user_message=message_to_send,  # Use resolved message
-                stream=True,
+                stream=False,
                 tags=[],
                 chat_session_id=self.session_id,
                 # Only include system instruction for model mode
@@ -155,36 +151,11 @@ class SendMessageCommand:
                 else None,
             )
 
-            async for chunk in self.client.generate_text_stream(request):
-                # Check for cancellation request
-                if self.store.state.cancel_requested:
-                    logger.info("stream_cancelled_by_user")
-                    was_cancelled = True
-                    return CommandResult[None](success=True, data=None)
-
-                # Debug: log chunk info for first few chunks
-                logger.debug(
-                    "stream_chunk",
-                    extra={
-                        "chunk_type": chunk.chunk_type,
-                        "has_content": bool(chunk.content),
-                        "has_reasoning": bool(chunk.reasoning_content),
-                        "sequence": chunk.sequence,
-                    },
-                )
-
-                # Handle thinking/reasoning chunks
-                if chunk.chunk_type == "thinking" and chunk.reasoning_content:
-                    if self.on_thinking_chunk:
-                        self.on_thinking_chunk(chunk.reasoning_content)
-                # Handle regular text content chunks
-                elif chunk.content:
-                    self.on_chunk(chunk.content)
+            await self.client.generate_text(request)
 
             return CommandResult[None](success=True, data=None)
 
         except Exception as e:
+            logger.warning("send_message_failed", error=str(e))
+            self.store.finish_generation()
             return CommandResult[None](success=False, error=str(e))
-
-        finally:
-            self.store.finish_streaming(cancelled=was_cancelled)

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/tui/chat/messages.py

@@ -89,44 +89,6 @@ class UndoRequested(Message):
     pass
 
 
-# =============================================================================
-# Streaming Events
-# =============================================================================
-
-
-class StreamingStarted(Message):
-    """Response streaming has begun."""
-
-    def __init__(self, execution_id: UUID) -> None:
-        self.execution_id = execution_id
-        super().__init__()
-
-
-class StreamingChunk(Message):
-    """A chunk of streamed content received."""
-
-    def __init__(self, content: str, is_reasoning: bool = False) -> None:
-        self.content = content
-        self.is_reasoning = is_reasoning
-        super().__init__()
-
-
-class StreamingCompleted(Message):
-    """Response streaming finished successfully."""
-
-    def __init__(self, full_content: str) -> None:
-        self.full_content = full_content
-        super().__init__()
-
-
-class StreamingError(Message):
-    """Error occurred during streaming."""
-
-    def __init__(self, error: str) -> None:
-        self.error = error
-        super().__init__()
-
-
 # =============================================================================
 # UI Events
 # =============================================================================

{alloy_runtime_cli-0.2.63 → alloy_runtime_cli-0.2.65}/cli/tui/chat/pane.py

@@ -25,7 +25,7 @@ class ChatPane(Widget):
 
     This widget contains the main chat interface components:
     - ChatHeader: Shows current session/model info
-    - MessageDisplay: Shows message history and streaming content
+    - MessageDisplay: Shows message history and pending request state
     - ChatInput: Text input for composing messages
 
     Each pane has its own ChatStore, allowing independent sessions