digitalkin 1.0.4.dev7__tar.gz → 1.0.5.dev2__tar.gz

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: digitalkin
-Version: 1.0.4.dev7
+Version: 1.0.5.dev2
 Summary: SDK to build kin used in DigitalKin
 Author-email: "DigitalKin.ai" <contact@digitalkin.ai>
 License: Attribution-NonCommercial-ShareAlike 4.0 International
@@ -453,7 +453,7 @@ Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: ag-ui-protocol==0.1.21.dev1787220377
-Requires-Dist: agentic-mesh-protocol==1.0.2
+Requires-Dist: agentic-mesh-protocol==1.0.1
 Requires-Dist: agno<3,>=2.8.0
 Requires-Dist: anyio>=4.13.0
 Requires-Dist: grpcio-health-checking==1.82.1

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/examples/services/filesystem_module.py

@@ -10,8 +10,7 @@ from pydantic import BaseModel, Field
 from digitalkin.logger import logger
 from digitalkin.models.module import ModuleStatus
 from digitalkin.modules.archetype_module import ArchetypeModule
-from digitalkin.models.services.filesystem import FileType
-from digitalkin.services.filesystem import FileFilter, UploadFileData
+from digitalkin.services.filesystem.filesystem_strategy import FileFilter, UploadFileData
 from digitalkin.services.services_config import ServicesConfig
 from digitalkin.services.services_models import ServicesMode
 
@@ -119,9 +118,9 @@ class ExampleModule(ArchetypeModule[ExampleInput, ExampleOutput, ExampleSetup, E
         file = UploadFileData(
             content=b"%s\n%s" % (processed_message.encode(), str(processed_number).encode()),
             name="example_output.txt",
-            type=FileType.DOCUMENT,
+            file_type="text/plain",
             content_type="text/plain",
-            metadata={"sdk_created": True},
+            metadata={"example_key": "example_value"},
             replace_if_exists=True,
         )
 
@@ -177,10 +176,10 @@ async def test_module() -> None:
     # Check the storage
     if module.status == ModuleStatus.STOPPED:
         files, _nb_results = await module.filesystem.get_files(
-            filters=FileFilter(names=["example_output.txt"]),
+            filters=FileFilter(name="example_output.txt", context="test-mission-123"),
         )
         for file in files:
-            await module.filesystem.update_file(file.id, type=FileType.OTHER)
+            await module.filesystem.update_file(file.id, file_type="updated")
             # module.filesystem.delete_files(filters=FileFilter(name="example_output.txt", context="test-mission-123"), permanent=True)
 
             logger.info("Retrieved file: %s with ID: %s", file.name, file.id)
@@ -189,7 +188,7 @@ async def test_module() -> None:
                 if file_record:
                     logger.info("File ID: %s", file_record.id)
                     logger.info("File name: %s", file_record.name)
-                    logger.info("File type: %s", file_record.type)
+                    logger.info("File type: %s", file_record.file_type)
                     logger.info("File status: %s", file_record.status)
                     logger.info("File content: %s", file_record.content.decode())
             except Exception:

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/pyproject.toml

@@ -31,7 +31,7 @@
         # which is how a nested run's output is attributed to its author. Pinned exactly
         # because it is a dated prerelease; move to >=0.1.21 once that ships stable.
         "ag-ui-protocol==0.1.21.dev1787220377",
-        "agentic-mesh-protocol==1.0.2",
+        "agentic-mesh-protocol==1.0.1",
         "agno>=2.8.0,<3",
         "anyio>=4.13.0",
         "grpcio-health-checking==1.82.1",
@@ -41,7 +41,7 @@
         "pydantic>=2.12.4",
         "redis[hiredis]>=7.4.0,<9",
     ]
-    version = "1.0.4.dev7"
+    version = "1.0.5.dev2"
 
     [project.optional-dependencies]
         agno = [ "agno>=2.6" ]
@@ -208,15 +208,15 @@
         ]
 
         ignore = [
-            "any-type",                       # Allow typing.Any — gRPC stubs, callbacks, and dynamic APIs require it
-            "missing-trailing-comma",         # Disable because of formatter incompatibility
-            "docstring-extraneous-exception", # Allow extraneous-exception in docstring
-            "unused-import",                  # Allow unused imports — used for availability checks
-            "invalid-function-name",          # Allow PascalCase methods — gRPC servicer convention
-            "import-outside-top-level",       # Allow lazy imports — needed to break circular dependencies
-            "bad-dunder-method-name",         # Allow __get_pydantic_core_schema__ — Pydantic dunder hook
-            "hardcoded-password-string",      # Hardcoded-password
-            "suspicious-pickle-import",       # Allow pickle import
+            "ANN401",  # Allow typing.Any — gRPC stubs, callbacks, and dynamic APIs require it
+            "COM812",  # Disable because of formatter incompatibility
+            "DOC502",  # Allow extraneous-exception in docstring
+            "F401",    # Allow unused imports — used for availability checks
+            "N802",    # Allow PascalCase methods — gRPC servicer convention
+            "PLC0415", # Allow lazy imports — needed to break circular dependencies
+            "PLW3201", # Allow __get_pydantic_core_schema__ — Pydantic dunder hook
+            "S105",    # Hardcoded-password
+            "S403",    # Allow pickle import
         ]
 
         fixable = [ "ALL" ]
@@ -226,40 +226,40 @@
 
         [tool.ruff.lint.per-file-ignores]
             "tests/**/*.py" = [
-                "missing-type-function-argument",
-                "missing-type-args",
-                "missing-type-kwargs",
-                "missing-return-type-undocumented-public-function",
-                "any-type",
-                "unused-function-argument",
-                "unused-method-argument",
-                "complex-structure",
-                "undocumented-public-module",
-                "undocumented-public-class",
-                "undocumented-public-method",
-                "undocumented-public-function",
-                "undocumented-public-init",
-                "undocumented-param",
-                "docstring-missing-returns",
-                "docstring-missing-exception",
-                "line-too-long",
-                "commented-out-code",
-                "boolean-type-hint-positional-argument",
-                "boolean-default-value-positional-argument",
-                "logging-f-string",
-                "invalid-function-name",
-                "try-except-in-loop",
-                "import-private-name",
-                "too-many-return-statements",
-                "too-many-branches",
-                "too-many-locals",
-                "too-many-statements",
-                "too-many-nested-blocks",
-                "magic-value-comparison",
-                "no-self-use",
-                "assert",
-                "private-member-access",
-                "raise-vanilla-class",
+                "ANN001",
+                "ANN002",
+                "ANN003",
+                "ANN201",
+                "ANN401",
+                "ARG001",
+                "ARG002",
+                "C901",
+                "D100",
+                "D101",
+                "D102",
+                "D103",
+                "D107",
+                "D417",
+                "DOC201",
+                "DOC501",
+                "E501",
+                "ERA001",
+                "FBT001",
+                "FBT002",
+                "G004",
+                "N802",
+                "PERF203",
+                "PLC2701",
+                "PLR0911",
+                "PLR0912",
+                "PLR0914",
+                "PLR0915",
+                "PLR1702",
+                "PLR2004",
+                "PLR6301",
+                "S101",
+                "SLF001",
+                "TRY002",
             ]
 
         [tool.ruff.lint.pylint]

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/src/digitalkin/__version__.py

@@ -5,4 +5,4 @@ from importlib.metadata import PackageNotFoundError, version
 try:
     __version__ = version("digitalkin")
 except PackageNotFoundError:
-    __version__ = "1.0.4.dev7"
+    __version__ = "1.0.5.dev2"

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/src/digitalkin/community/agno/agno_adapter.py

@@ -7,6 +7,8 @@ import logging
 import uuid
 from typing import TYPE_CHECKING, Any, TypeAlias
 
+from pydantic import ValidationError
+
 if TYPE_CHECKING:
     from collections.abc import Callable
 
@@ -54,6 +56,7 @@ if TYPE_CHECKING:
 from digitalkin.models.events import (
     AgentRunEvent,
     BaseAgentRunEvent,
+    CustomEvent,
     ReasoningCompletedEvent,
     ReasoningContentDeltaEvent,
     ReasoningStartedEvent,
@@ -62,6 +65,8 @@ from digitalkin.models.events import (
     RunContentEvent,
     RunErrorEvent,
     RunStartedEvent,
+    SourceCitation,
+    SourceCitationEvent,
     SubagentErrorEvent,
     SubagentFinishedEvent,
     SubagentStartedEvent,
@@ -80,7 +85,9 @@ class AgnoStreamAdapter:
     """Stateful Agno→DigitalKin event converter.
 
     Auto-wraps ``run_content`` deltas in TextMessage/Reasoning lifecycle
-    events and tracks HITL pause state.
+    events and tracks HITL pause state. One adapter serves one stream: its
+    dedup state (citations, closed tool calls, completed runs) is reset on each
+    top-level ``run_started``, so it must not be shared across concurrent streams.
     """
 
     def __init__(self) -> None:
@@ -94,6 +101,8 @@ class AgnoStreamAdapter:
         self._reasonings: dict[str, tuple[str, dict[str, Any] | None]] = {}
 
         self._closed_tool_call_ids: set[str] = set()
+        # Providers repeat the same citations on every delta and again on completion.
+        self._cited_urls: set[str] = set()
 
         self._active_run_id: str | None = None
         self._completed_run_ids: set[str] = set()
@@ -206,6 +215,7 @@ class AgnoStreamAdapter:
             "tool_call_started": self._handle_tool_call_started,
             "tool_call_completed": self._handle_tool_call_completed,
             "tool_call_error": self._handle_tool_call_error,
+            "custom_event": self._handle_custom_event,
         }
         dispatch = {
             enum_cls[name]: handler
@@ -313,6 +323,17 @@ class AgnoStreamAdapter:
             self._active_run_id,
             self._last_metadata,
         )
+        # TODO(validate): ADAPTER-RESET per-run dedup state resets on each new top-level run
+        logger.info(
+            "[VALIDATE ADAPTER-RESET] run_id=%s clearing cited_urls=%d closed_tool_calls=%d completed_runs=%d",
+            run_id,
+            len(self._cited_urls),
+            len(self._closed_tool_call_ids),
+            len(self._completed_run_ids),
+        )
+        self._cited_urls.clear()
+        self._closed_tool_call_ids.clear()
+        self._completed_run_ids.clear()
         self._active_run_id = run_id
         return [
             RunStartedEvent(
@@ -337,11 +358,13 @@ class AgnoStreamAdapter:
         """
         parent_run_id = getattr(agno_event, "parent_run_id", None)
         run_id = agno_event.run_id
+        citations = self._citations(agno_event, timestamp)
 
         if parent_run_id:
             # Close this member's own text/reasoning bubble, then the matching step. Siblings
             # still streaming keep theirs open.
-            events: list[BaseAgentRunEvent] = self._close_content(self._run_key, timestamp)
+            events: list[BaseAgentRunEvent] = citations
+            events.extend(self._close_content(self._run_key, timestamp))
             events.extend(self._close_reasoning(self._run_key, timestamp))
 
             subagent = self._subagents.pop(run_id, None) if run_id else None
@@ -377,7 +400,8 @@ class AgnoStreamAdapter:
 
         # AG-UI refuses RUN_FINISHED while any message, reasoning or step is still open, so
         # everything still running when the top-level run ends is force-closed here.
-        events = self._close_all_content(timestamp)
+        events = citations
+        events.extend(self._close_all_content(timestamp))
         events.extend(self._close_all_reasoning(timestamp))
         events.extend(self._close_subagents(timestamp))
 
@@ -884,6 +908,80 @@ class AgnoStreamAdapter:
             event for run_key in reversed(list(self._messages)) for event in self._close_content(run_key, timestamp)
         ]
 
+    def _source_event(self, payload: Any, timestamp: Any) -> SourceCitationEvent | None:
+        """Validate one source payload into a ``source_citation`` event.
+
+        Args:
+            payload: ``{url, title?, description?}`` as received.
+            timestamp: Event timestamp.
+
+        Returns:
+            The event, or None when the payload is invalid (logged and dropped).
+        """
+        try:
+            citation = SourceCitation.model_validate(payload)
+        except ValidationError as exc:
+            logger.warning("Dropped source citation %r: %s", payload, exc)
+            return None
+        return SourceCitationEvent(
+            event=AgentRunEvent.CUSTOM,
+            name="source_citation",
+            value=citation,
+            subagent_run_id=self._subagent_of(self._run_key),
+            timestamp=timestamp,
+            metadata=self._last_metadata,
+        )
+
+    def _citations(
+        self, agno_event: AgnoRunContentEvent | AgnoRunCompletedEvent, timestamp: Any
+    ) -> list[BaseAgentRunEvent]:
+        """Convert the model-native ``citations.urls`` not yet emitted in this run.
+
+        Returns:
+            One SourceCitationEvent per new URL.
+        """
+        citations = agno_event.citations
+        if citations is None or not citations.urls:
+            return []
+        events: list[BaseAgentRunEvent] = []
+        for cite in citations.urls:
+            if not cite.url:
+                logger.warning("Dropped source citation without url: title=%r", cite.title)
+                continue
+            if cite.url in self._cited_urls:
+                continue
+            self._cited_urls.add(cite.url)
+            event = self._source_event({"url": cite.url, "title": cite.title}, timestamp)
+            if event is not None:
+                events.append(event)
+        return events
+
+    def _handle_custom_event(self, agno_event: AgnoRunEvent, timestamp: Any) -> list[BaseAgentRunEvent]:
+        """Handle RunEvent.custom_event — agno stores the tool's kwargs as attributes.
+
+        Returns:
+            A SourceCitationEvent for ``source_citation``, a generic CustomEvent otherwise,
+            or empty when the event is unnamed or its source is invalid.
+        """
+        data = agno_event.__dict__
+        name = data.get("name")
+        if not name:
+            logger.debug("Skipping unnamed agno custom event: %r", data)
+            return []
+        if name == "source_citation":
+            event = self._source_event(data.get("value"), timestamp)
+            return [event] if event is not None else []
+        return [
+            CustomEvent(
+                event=AgentRunEvent.CUSTOM,
+                name=name,
+                value=data.get("value"),
+                subagent_run_id=self._subagent_of(self._run_key),
+                timestamp=timestamp,
+                metadata=self._last_metadata,
+            )
+        ]
+
     def _handle_run_content(self, agno_event: AgnoRunContentEvent, timestamp: Any) -> list[BaseAgentRunEvent]:
         """Handle RunEvent.run_content — the core state machine.
 
@@ -892,7 +990,7 @@ class AgnoStreamAdapter:
         Returns:
             DigitalKin events for this chunk.
         """
-        events: list[BaseAgentRunEvent] = []
+        events = self._citations(agno_event, timestamp)
 
         reasoning_content = agno_event.reasoning_content
         content = agno_event.content

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/src/digitalkin/community/agno/hitl.py

@@ -85,11 +85,26 @@ class PausedRunStore:
             if tid and tid not in seen and tool.external_execution_required and tool.result is None:
                 seen.add(tid)
                 pending.append(tid)
+        payload = run_output.to_dict()
+        # History messages are re-fetched from the session by continue_run; events are not replayed.
+        messages = payload.get("messages") or []
+        history = [m for m in messages if m.get("from_history")]
+        if history:
+            payload["messages"] = [m for m in messages if not m.get("from_history")]
+        events = payload.pop("events", None) or []
+        # TODO(validate): HITL-TRIM paused runs stored without history messages/events still resume
+        logger.info(
+            "[VALIDATE HITL-TRIM] thread_id=%s dropped %d history message(s) %s and %d event(s)",
+            thread_id,
+            len(history),
+            [(m.get("id"), m.get("role")) for m in history],
+            len(events),
+        )
         record = PausedRunRecord(
             thread_id=thread_id,
             run_id=run_output.run_id or "",
             pending_tool_call_ids=pending,
-            payload=run_output.to_dict(),
+            payload=payload,
         )
         await self._storage.upsert(
             collection=self.COLLECTION,

{digitalkin-1.0.4.dev7 → digitalkin-1.0.5.dev2}/src/digitalkin/community/agno/models.py

@@ -67,8 +67,6 @@ class ToolCallMetadata(BaseModel):
         duration_ms: Execution time in milliseconds.
         cost_tracked: Whether cost metadata is available from the tool.
         error: Error message if the call failed, None otherwise.
-        input_kwargs: The kwargs passed to the tool (for debugging).
-        output_summary: Brief summary of the output (truncated for logs).
         tool_metadata: Metadata returned by the tool itself (cost, API calls, etc.).
     """
 
@@ -77,14 +75,6 @@ class ToolCallMetadata(BaseModel):
     duration_ms: float = Field(..., description="Execution time in milliseconds")
     cost_tracked: bool = Field(default=False, description="Whether cost metadata is available from tool")
     error: str | None = Field(default=None, description="Error message if failed")
-    input_kwargs: dict[str, Any] | None = Field(
-        default=None,
-        description="Input kwargs passed to the tool for debugging",
-    )
-    output_summary: str | None = Field(
-        default=None,
-        description="Brief summary of the output (truncated)",
-    )
     tool_metadata: ToolOutputMetadata | None = Field(
         default=None,
         description="Metadata returned by the tool module (cost, API calls, etc.)",
@@ -94,7 +84,7 @@ class ToolCallMetadata(BaseModel):
         """Convert metadata to dict for successful responses.
 
         Returns:
-            Dictionary with success-relevant fields (error excluded).
+            Dictionary with success-relevant fields; error is excluded.
         """
         return self.model_dump(exclude={"error"})
 
@@ -106,31 +96,6 @@ class ToolCallMetadata(BaseModel):
         """
         return self.model_dump(include={"module_id", "success", "duration_ms", "error"})
 
-    def to_log_dict(self) -> dict[str, Any]:
-        """Convert metadata to dict for logging.
-
-        Returns:
-            Dictionary suitable for log extra data, large kwargs truncated.
-        """
-        data: dict[str, Any] = {
-            "module_id": self.module_id,
-            "success": self.success,
-            "duration_ms": self.duration_ms,
-            "cost_tracked": self.cost_tracked,
-        }
-        if self.error:
-            data["error"] = self.error
-        if self.input_kwargs:
-            max_len = 100
-            data["input_kwargs"] = {
-                k: (str(v)[:max_len] + "..." if len(str(v)) > max_len else v) for k, v in self.input_kwargs.items()
-            }
-        if self.output_summary:
-            data["output_summary"] = self.output_summary
-        if self.tool_metadata:
-            data["tool_metadata"] = self.tool_metadata.model_dump(exclude_none=True)
-        return data
-
     @staticmethod
     def extract_tool_metadata(output: dict[str, Any]) -> "ToolOutputMetadata | None":
         """Extract tool metadata from a tool's output if present.