fastapi-fullstack 0.2.15__tar.gz → 0.2.16__tar.gz

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fastapi-fullstack
-Version: 0.2.15
+Version: 0.2.16
 Summary: Full-stack FastAPI + Next.js template generator with PydanticAI/LangChain agents, WebSocket streaming, 20+ enterprise integrations, and Logfire/LangSmith observability. Ship AI apps fast. CLI tool to generate production-ready FastAPI + Next.js projects with AI agents, auth, and observability.
 Project-URL: Homepage, https://github.com/vstorm-co/full-stack-ai-agent-template
 Project-URL: Documentation, https://github.com/vstorm-co/full-stack-ai-agent-template#readme
@@ -21,7 +21,7 @@ Classifier: Topic :: Software Development :: Code Generators
 Requires-Python: >=3.11
 Requires-Dist: arq>=0.28.0
 Requires-Dist: celery>=5.6.3
-Requires-Dist: click>=8.3.0
+Requires-Dist: click>=8.3.3
 Requires-Dist: cookiecutter>=2.7.0
 Requires-Dist: email-validator>=2.3.0
 Requires-Dist: httpx>=0.28.1

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/fastapi_gen/cli.py

@@ -50,6 +50,7 @@ def _preflight_check(  # noqa: C901
     frontend: str,
     admin_panel: bool,
     marketing_site: bool,
+    demo_export: bool,
     oauth_google: bool,
     gdrive_rag: bool,
     s3_rag: bool,
@@ -152,6 +153,13 @@ def _preflight_check(  # noqa: C901
                 "Add --frontend nextjs or drop --marketing-site",
             )
         )
+    if demo_export and frontend == "none":
+        issues.append(
+            (
+                "--demo-export requires --frontend nextjs (it bundles the real replay UI into the HTML)",
+                "Add --frontend nextjs or drop --demo-export",
+            )
+        )
     if admin_panel and frontend == "none":
         click.echo(
             click.style(
@@ -631,6 +639,12 @@ def new(output: Path | None, no_input: bool, name: str | None, minimal: bool) ->
     default=False,
     help="Generate marketing/landing pages",
 )
+@click.option(
+    "--demo-export",
+    is_flag=True,
+    default=False,
+    help="Self-contained HTML demo export (Vite bundle + exporter script; requires --frontend)",
+)
 @click.option(
     "--i18n/--no-i18n",
     "i18n",
@@ -793,6 +807,7 @@ def create(
     email_provider: str,
     newsletter: bool,
     marketing_site: bool,
+    demo_export: bool,
     i18n: bool,
     example_resource: bool,
     changelog: bool,
@@ -1117,6 +1132,7 @@ def create(
                 frontend=frontend,
                 admin_panel=admin_panel,
                 marketing_site=marketing_site,
+                demo_export=demo_export,
                 oauth_google=oauth_google,
                 gdrive_rag=gdrive_rag,
                 s3_rag=s3_rag,
@@ -1213,6 +1229,7 @@ def create(
                 email_provider=EmailProviderType(email_provider),
                 enable_newsletter_signup=newsletter,
                 enable_marketing_site=marketing_site,
+                enable_demo_export=demo_export,
                 enable_i18n=i18n,
                 include_example_crud=example_resource,
                 enable_changelog=changelog,

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/fastapi_gen/config.py

@@ -395,6 +395,7 @@ class ProjectConfig(BaseModel):
 
     # Marketing / Frontend pages
     enable_marketing_site: bool = False
+    enable_demo_export: bool = False
     enable_i18n: bool = True
     include_example_crud: bool = False
     enable_changelog: bool = False
@@ -618,6 +619,11 @@ class ProjectConfig(BaseModel):
                 "Marketing site requires a frontend (landing/blog/legal pages need a UI). "
                 "Quick fix: add --frontend nextjs, or drop --marketing-site."
             )
+        if self.enable_demo_export and self.frontend == FrontendType.NONE:
+            raise ValueError(
+                "Demo export requires a frontend (it bundles the real replay UI into the HTML). "
+                "Quick fix: add --frontend nextjs, or drop --demo-export."
+            )
         if self.oauth_provider != OAuthProvider.NONE and self.frontend == FrontendType.NONE:
             raise ValueError(
                 "OAuth requires a frontend for the callback page. "
@@ -969,6 +975,7 @@ class ProjectConfig(BaseModel):
             "enable_admin_features_system_health": self.enable_admin_features_system_health,
             # Marketing / Frontend pages
             "enable_marketing_site": self.enable_marketing_site,
+            "enable_demo_export": self.enable_demo_export and self.frontend != FrontendType.NONE,
             "enable_changelog": self.enable_changelog,
             "enable_testimonials": self.enable_testimonials,
             "enable_comparison_pages": self.enable_comparison_pages,

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/fastapi_gen/prompts.py

@@ -1450,6 +1450,11 @@ def prompt_marketing_features() -> dict[str, bool]:
                     value="status_badge",
                     checked=True,
                 ),
+                questionary.Choice(
+                    "Demo export (offline HTML replay of a saved conversation)",
+                    value="demo_export",
+                    checked=False,
+                ),
             ],
         ).ask()
     )
@@ -1461,6 +1466,7 @@ def prompt_marketing_features() -> dict[str, bool]:
         "enable_comparison_pages": "comparison_pages" in features,
         "enable_affiliate_program": "affiliate_program" in features,
         "enable_status_badge": "status_badge" in features,
+        "enable_demo_export": "demo_export" in features,
     }
 
 
@@ -1801,6 +1807,7 @@ def run_interactive_prompts() -> ProjectConfig:
             "enable_comparison_pages": True,
             "enable_affiliate_program": True,
             "enable_status_badge": True,
+            "enable_demo_export": False,
         }
         if frontend != FrontendType.NONE
         else {}

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "fastapi-fullstack"
-version = "0.2.15"
+version = "0.2.16"
 description = "Full-stack FastAPI + Next.js template generator with PydanticAI/LangChain agents, WebSocket streaming, 20+ enterprise integrations, and Logfire/LangSmith observability. Ship AI apps fast. CLI tool to generate production-ready FastAPI + Next.js projects with AI agents, auth, and observability."
 readme = "README.md"
 requires-python = ">=3.11"
@@ -44,7 +44,7 @@ classifiers = [
 ]
 
 dependencies = [
-    "click>=8.3.0",
+    "click>=8.3.3",
     "cookiecutter>=2.7.0",
     "rich>=15.0.0",
     "questionary>=2.1.0",

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/VARIABLES.md

@@ -196,6 +196,7 @@ These variables are set automatically by the generator.
 | `enable_cors` | bool | `true` | Enable CORS middleware | - |
 | `enable_webhooks` | bool | `false` | Enable webhook support | - |
 | `enable_conversation_persistence` | bool | `true` | Enable conversation persistence (derived from `use_ai`) | Computed from `use_ai` |
+| `enable_demo_export` | bool | `false` | Self-contained HTML demo export: a `frontend/demo-export/` Vite single-file bundle + `scripts/export_demo_html.py` that render a saved conversation to one offline `.html` replay | Requires a frontend |
 | `include_example_crud` | bool | `false` | Include example CRUD endpoints (always disabled) | Always false |
 | `enable_i18n` | bool | `true` | Enable internationalization in frontend (always enabled) | Always true |
 | `seed_admin_email` | string | `""` | Email of a user to auto-promote to app-admin on first startup (via `FIRST_ADMIN_EMAIL` env var) | - |

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/cookiecutter.json

@@ -84,6 +84,7 @@
   "use_all_providers": false,
   "use_ai": true,
   "enable_conversation_persistence": true,
+  "enable_demo_export": false,
   "enable_langsmith": false,
   "enable_web_search": false,
   "enable_web_fetch": false,

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/hooks/post_gen_project.py

@@ -41,7 +41,9 @@ enable_rag_image_description = "{{ cookiecutter.enable_rag_image_description }}"
 enable_google_drive_ingestion = "{{ cookiecutter.enable_google_drive_ingestion }}" == "True"
 enable_s3_ingestion = "{{ cookiecutter.enable_s3_ingestion }}" == "True"
 enable_web_search = "{{ cookiecutter.enable_web_search }}" == "True"
+enable_web_fetch = "{{ cookiecutter.enable_web_fetch }}" == "True"
 web_fetch_tool = "{{ cookiecutter.web_fetch_tool }}" == "True"
+enable_demo_export = "{{ cookiecutter.enable_demo_export }}" == "True"
 enable_charts = "{{ cookiecutter.enable_charts }}" == "True"
 charts_channel_png = "{{ cookiecutter.charts_channel_png }}" == "True"
 enable_code_execution = "{{ cookiecutter.enable_code_execution }}" == "True"
@@ -154,6 +156,20 @@ if not enable_deep_research and use_frontend:
     frontend_src = os.path.join(os.getcwd(), "frontend", "src")
     remove_file(os.path.join(frontend_src, "stores", "research-store.ts"))
     remove_file(os.path.join(frontend_src, "stores", "chat-mode-store.ts"))
+    remove_file(os.path.join(frontend_src, "lib", "research-from-tools.ts"))
+    remove_file(os.path.join(frontend_src, "components", "chat", "research-replay-block.tsx"))
+
+# The fetched-page tool-result renderer is only referenced when web fetch is on.
+if not enable_web_fetch and use_frontend:
+    frontend_src = os.path.join(os.getcwd(), "frontend", "src")
+    remove_file(os.path.join(frontend_src, "components", "chat", "tool-results", "fetch-url.tsx"))
+
+# Self-contained HTML demo export (Vite single-file bundle + exporter script).
+# Only meaningful with a frontend to bundle; drop everything otherwise. The
+# scripts/ dir exists solely for this exporter, so remove it wholesale.
+if not (enable_demo_export and use_frontend):
+    remove_dir(os.path.join(os.getcwd(), "frontend", "demo-export"))
+    remove_dir(os.path.join(os.getcwd(), "scripts"))
 
 if not enable_subagents and use_frontend:
     frontend_src = os.path.join(os.getcwd(), "frontend", "src")

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/.gitignore

@@ -99,5 +99,10 @@ logs/
 # OS
 .DS_Store
 Thumbs.db
+{%- if cookiecutter.enable_demo_export %}
+
+# Demo-export Vite build output (regenerated by `bun run build:demo-export`)
+frontend/demo-export/dist/
+{%- endif %}
 
 # Project specific

fastapi_fullstack-0.2.16/template/{{cookiecutter.project_slug}}/CLAUDE.md

@@ -0,0 +1,92 @@
+# CLAUDE.md
+
+## Project Overview
+
+**{{ cookiecutter.project_name }}** - FastAPI application generated with [Full-Stack AI Agent Template](https://github.com/vstorm-co/full-stack-ai-agent-template).
+
+**Stack:** FastAPI + Pydantic v2
+{%- if True %}, PostgreSQL (async via asyncpg){%- endif %}
+{%- if False %}, MongoDB (async via Motor){%- endif %}
+{%- if False %}, SQLite (sync){%- endif %}
+, JWT + API Key auth
+{%- if cookiecutter.enable_redis %}, Redis{%- endif %}
+{%- if cookiecutter.use_pydantic_ai %}, PydanticAI{%- endif %}
+{%- if cookiecutter.use_langchain %}, LangChain{%- endif %}
+{%- if cookiecutter.use_langgraph %}, LangGraph{%- endif %}{%- if cookiecutter.use_deepagents %}, DeepAgents{%- endif %}
+{%- if cookiecutter.enable_rag %}, RAG ({{ cookiecutter.vector_store }}){%- endif %}
+{%- if cookiecutter.use_celery %}, Celery{%- endif %}
+{%- if cookiecutter.use_taskiq %}, Taskiq{%- endif %}
+{%- if cookiecutter.use_frontend %}, Next.js 15 (i18n){%- endif %}
+
+## Commands
+
+```bash
+# Backend
+cd backend
+uv run uvicorn app.main:app --reload --port {{ cookiecutter.backend_port }}
+uv run pytest
+uv run pytest tests/test_file.py::test_name -v
+uv run ruff check . --fix && uv run ruff format .
+uv run ty check
+
+# Database migrations
+uv run alembic upgrade head
+uv run alembic revision --autogenerate -m "Description"
+{%- if cookiecutter.use_frontend %}
+
+# Frontend
+cd frontend
+bun dev
+bun test
+bun run lint
+{%- endif %}
+{%- if cookiecutter.enable_docker %}
+
+# Docker
+docker compose up -d
+{%- endif %}
+{%- if cookiecutter.enable_rag %}
+
+# RAG
+uv run {{ cookiecutter.project_slug }} rag-collections
+uv run {{ cookiecutter.project_slug }} rag-ingest /path/to/file.pdf --collection docs
+uv run {{ cookiecutter.project_slug }} rag-search "query" --collection docs
+{%- if cookiecutter.enable_google_drive_ingestion %}
+uv run {{ cookiecutter.project_slug }} rag-sync-gdrive --collection docs
+{%- endif %}
+{%- if cookiecutter.enable_s3_ingestion %}
+uv run {{ cookiecutter.project_slug }} rag-sync-s3 --collection docs
+{%- endif %}
+
+# Sync Sources
+uv run {{ cookiecutter.project_slug }} cmd rag-sources
+uv run {{ cookiecutter.project_slug }} cmd rag-source-add
+uv run {{ cookiecutter.project_slug }} cmd rag-source-sync
+{%- endif %}
+```
+
+## Hard Boundaries
+
+Non-obvious rules that are easy to violate and cross-cutting enough to state up front:
+
+- Repositories use `db.flush()` + `db.refresh()`, **never** `db.commit()` — the session auto-commits via `get_db_session`.
+- Routes call services only — **never** import or call repositories directly.
+- Route handlers return `-> Any`; serialization is handled by `response_model` (avoids double Pydantic validation).
+- `datetime.now(UTC)`, never `datetime.utcnow()`.
+- `secrets.compare_digest()` for API key comparison, never `==`.
+
+## Detailed Conventions
+
+Path-scoped guidance lives in `.claude/rules/*` and loads automatically when you edit matching files — it is intentionally NOT repeated here:
+
+- `architecture.md` — Routes → Services → Repositories, dependency injection, thin vs. thick domains
+- `schemas-models.md` — Pydantic v2 schemas (`*Create`/`*Update`/`*Read`/`*List`), SQLAlchemy models
+- `api-conventions.md` — REST structure, status codes, response format, pagination, auth
+- `exceptions-security.md` — domain exceptions (`NotFoundError`, etc.), JWT, RBAC
+- `code-style.md` — formatting, naming, imports, type hints
+- `testing.md` — test structure, fixtures, async patterns
+{%- if cookiecutter.use_frontend %}
+- `frontend.md` — Next.js 15 conventions
+{%- endif %}
+
+Longer-form docs: `docs/architecture.md`, `docs/adding_features.md`, `docs/testing.md`, `docs/patterns.md`.

fastapi_fullstack-0.2.16/template/{{cookiecutter.project_slug}}/backend/alembic/versions/0025_message_thinking.py

@@ -0,0 +1,27 @@
+"""messages: add thinking column for the reasoning trace
+
+Revision ID: 0025
+Revises: 0024_create_webhook_tables
+
+Adds a nullable ``thinking`` text column to messages so the assistant's
+reasoning trace (streamed live as ``thinking_delta``) is persisted alongside
+the turn. Enables loaded conversations — and the self-contained HTML export —
+to show the THINKING block, not just live streaming.
+"""
+
+import sqlalchemy as sa
+
+from alembic import op
+
+revision = "0025"
+down_revision = "0024_create_webhook_tables"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column("messages", sa.Column("thinking", sa.Text(), nullable=True))
+
+
+def downgrade() -> None:
+    op.drop_column("messages", "thinking")

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/app/db/models/conversation.py

@@ -113,6 +113,7 @@ class Message(TimestampMixin, SQLModel, table=True):
     )
     role: str = Field(max_length=20)  # user, assistant, system
     content: str = Field(sa_column=Column(Text, nullable=False))
+    thinking: str | None = Field(default=None, sa_column=Column(Text, nullable=True))
     model_name: str | None = Field(default=None, max_length=100)
     tokens_used: int | None = Field(default=None)
 
@@ -291,6 +292,7 @@ class Message(Base, TimestampMixin):
         String(20), nullable=False
     )  # user, assistant, system
     content: Mapped[str] = mapped_column(Text, nullable=False)
+    thinking: Mapped[str | None] = mapped_column(Text, nullable=True)
     model_name: Mapped[str | None] = mapped_column(String(100), nullable=True)
     tokens_used: Mapped[int | None] = mapped_column(Integer, nullable=True)
 

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/app/repositories/conversation.py

@@ -411,6 +411,7 @@ async def create_message(
     conversation_id: UUID,
     role: str,
     content: str,
+    thinking: str | None = None,
     model_name: str | None = None,
     tokens_used: int | None = None,
 ) -> Message:
@@ -419,6 +420,7 @@ async def create_message(
         conversation_id=conversation_id,
         role=role,
         content=content,
+        thinking=thinking,
         model_name=model_name,
         tokens_used=tokens_used,
     )

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/app/schemas/conversation.py

@@ -72,6 +72,7 @@ class MessageBase(BaseSchema):
 
     role: Literal["user", "assistant", "system"] = Field(..., description="Message role")
     content: str = Field(..., description="Message content")
+    thinking: str | None = Field(default=None, description="Reasoning trace (assistant turns)")
 
 
 class MessageCreate(MessageBase):

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/app/services/agent.py

@@ -223,6 +223,7 @@ async def persist_assistant_turn(
     output: str,
     model_name: str | None,
     collected_tool_calls: list[dict[str, Any]],
+    thinking: str | None = None,
 ) -> str | None:
     """Persist the assistant message and any tool calls. Returns the saved message id."""
     try:
@@ -230,7 +231,9 @@ async def persist_assistant_turn(
             conv_service = get_conversation_service(db)
             assistant_msg = await conv_service.add_message(
                 UUID(conversation_id),
-                MessageCreate(role="assistant", content=output, model_name=model_name),
+                MessageCreate(
+                    role="assistant", content=output, thinking=thinking, model_name=model_name
+                ),
             )
             for tc in collected_tool_calls:
                 try:

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/app/services/agent_session.py

@@ -250,6 +250,7 @@ class AgentSession:
 {%- endif %}
 
             collected_tool_calls: list[dict[str, Any]] = []
+            collected_thinking: list[str] = []
 {%- if cookiecutter.enable_deep_research %}
 {%- if cookiecutter.enable_subagents %}
             self._subagent_task_manager = (
@@ -277,7 +278,9 @@ class AgentSession:
                 async with assistant.agent.iter(
                     user_input, deps=self.deps, message_history=model_history
                 ) as agent_run:
-                    await self._stream_agent_run(agent_run, user_message, collected_tool_calls)
+                    await self._stream_agent_run(
+                        agent_run, user_message, collected_tool_calls, collected_thinking
+                    )
             finally:
                 if poller is not None:
                     poller.cancel()
@@ -296,7 +299,9 @@ class AgentSession:
             async with assistant.agent.iter(
                 user_input, deps=self.deps, message_history=model_history
             ) as agent_run:
-                await self._stream_agent_run(agent_run, user_message, collected_tool_calls)
+                await self._stream_agent_run(
+                    agent_run, user_message, collected_tool_calls, collected_thinking
+                )
 {%- endif %}
 
             # Update in-memory history only after a complete agent run
@@ -314,6 +319,7 @@ class AgentSession:
                     agent_run.result.output,
                     getattr(assistant, "model_name", None),
                     collected_tool_calls,
+                    thinking="".join(collected_thinking) or None,
                 )
 
 {%- if cookiecutter.enable_billing and cookiecutter.enable_teams and cookiecutter.enable_credits_system %}
@@ -601,6 +607,7 @@ class AgentSession:
         agent_run: Any,
         user_message: str,
         collected_tool_calls: list[dict[str, Any]],
+        collected_thinking: list[str],
     ) -> None:
         """Drive the agent_run iterator, dispatching each node to its streaming helper."""
         async for node in agent_run:
@@ -614,7 +621,7 @@ class AgentSession:
             elif Agent.is_model_request_node(node):
                 await send_event(self.websocket, "model_request_start", {})
                 async with node.stream(agent_run.ctx) as request_stream:
-                    await self._stream_request_events(request_stream)
+                    await self._stream_request_events(request_stream, collected_thinking)
             elif Agent.is_call_tools_node(node):
                 await send_event(self.websocket, "call_tools_start", {})
                 async with node.stream(agent_run.ctx) as handle_stream:
@@ -624,7 +631,9 @@ class AgentSession:
                     self.websocket, "final_result", {"output": agent_run.result.output}
                 )
 
-    async def _stream_request_events(self, request_stream: Any) -> None:
+    async def _stream_request_events(
+        self, request_stream: Any, collected_thinking: list[str]
+    ) -> None:
         """Forward model-request events (text/thinking/tool deltas + final-result start).
 {%- if cookiecutter.enable_deep_research %}
 
@@ -673,6 +682,9 @@ class AgentSession:
                     )
 {%- endif %}
                 elif isinstance(event.part, ThinkingPart) and event.part.content:
+                    if collected_thinking:
+                        collected_thinking.append(" ")
+                    collected_thinking.append(event.part.content)
                     await send_event(
                         self.websocket,
                         "thinking_delta",
@@ -691,6 +703,7 @@ class AgentSession:
 {%- endif %}
                 elif isinstance(event.delta, ThinkingPartDelta):
                     if event.delta.content_delta:
+                        collected_thinking.append(event.delta.content_delta)
                         await send_event(
                             self.websocket,
                             "thinking_delta",
@@ -814,6 +827,10 @@ class AgentSession:
         self.current_conversation_id: str | None = None
 {%- endif %}
         self._last_usage_metadata: Any = None
+        # TODO: reasoning is streamed to the UI (thinking_delta) but not collected for
+        # persistence here — Message.thinking stays empty for this framework. The text is
+        # already extracted below; accumulate it per turn and pass it to
+        # persist_assistant_turn(thinking=...) like the PydanticAI session does.
         self._thinking_streamed: bool = False
         self._turn_task: asyncio.Task[None] | None = None
 
@@ -1274,6 +1291,10 @@ class AgentSession:
         self.current_conversation_id: str | None = None
 {%- endif %}
         self._last_usage_metadata: Any = None
+        # TODO: reasoning is streamed to the UI (thinking_delta) but not collected for
+        # persistence here — Message.thinking stays empty for this framework. The text is
+        # already extracted below; accumulate it per turn and pass it to
+        # persist_assistant_turn(thinking=...) like the PydanticAI session does.
         self._thinking_streamed: bool = False
         self._turn_task: asyncio.Task[None] | None = None
 
@@ -1743,6 +1764,10 @@ class AgentSession:
         # invalidate any pending interrupt anyway).
         self._current_thinking_effort: str | None = None
         self._last_usage_metadata: Any = None
+        # TODO: reasoning is streamed to the UI (thinking_delta) but not collected for
+        # persistence here — Message.thinking stays empty for this framework. The text is
+        # already extracted below; accumulate it per turn and pass it to
+        # persist_assistant_turn(thinking=...) like the PydanticAI session does.
         self._thinking_streamed: bool = False
 {%- if cookiecutter.use_database %}
         self.current_conversation_id: str | None = None
@@ -2465,17 +2490,19 @@ class AgentSession:
             kb_token = _active_kb_collections.set(kb_names)
             try:
                 collected_tool_calls: list[dict[str, Any]] = []
+                collected_thinking: list[str] = []
                 async with assistant.agent.iter(user_input, deps=assistant.deps) as agent_run:
                     await self._stream_agent_run(
-                        agent_run, user_message, collected_tool_calls
+                        agent_run, user_message, collected_tool_calls, collected_thinking
                     )
             finally:
                 _active_kb_collections.reset(kb_token)
 {%- else %}
             collected_tool_calls: list[dict[str, Any]] = []
+            collected_thinking: list[str] = []
             async with assistant.agent.iter(user_input, deps=assistant.deps) as agent_run:
                 await self._stream_agent_run(
-                    agent_run, user_message, collected_tool_calls
+                    agent_run, user_message, collected_tool_calls, collected_thinking
                 )
 {%- endif %}
 
@@ -2486,6 +2513,7 @@ class AgentSession:
                     agent_run.result.output,
                     getattr(assistant, "model_name", None),
                     collected_tool_calls,
+                    thinking="".join(collected_thinking) or None,
                 )
 
             await send_event(
@@ -2587,6 +2615,7 @@ class AgentSession:
         agent_run: Any,
         user_message: str,
         collected_tool_calls: list[dict[str, Any]],
+        collected_thinking: list[str],
     ) -> None:
         """Drive the pydantic-ai agent_run iterator, forwarding all events."""
         async for node in agent_run:
@@ -2600,7 +2629,7 @@ class AgentSession:
             elif Agent.is_model_request_node(node):
                 await send_event(self.websocket, "model_request_start", {})
                 async with node.stream(agent_run.ctx) as request_stream:
-                    await self._stream_request_events(request_stream)
+                    await self._stream_request_events(request_stream, collected_thinking)
             elif Agent.is_call_tools_node(node):
                 await send_event(self.websocket, "call_tools_start", {})
                 async with node.stream(agent_run.ctx) as handle_stream:
@@ -2610,7 +2639,9 @@ class AgentSession:
                     self.websocket, "final_result", {"output": agent_run.result.output}
                 )
 
-    async def _stream_request_events(self, request_stream: Any) -> None:
+    async def _stream_request_events(
+        self, request_stream: Any, collected_thinking: list[str]
+    ) -> None:
         """Forward model-request events (text/thinking/tool deltas + final-result start)."""
         async for event in request_stream:
             if isinstance(event, PartStartEvent):
@@ -2628,6 +2659,9 @@ class AgentSession:
                 elif isinstance(event.part, ThinkingPart) and event.part.content:
                     # Surface the model's reasoning trace to the UI. Anthropic +
                     # OpenAI-reasoning models emit these as the model "thinks".
+                    if collected_thinking:
+                        collected_thinking.append(" ")
+                    collected_thinking.append(event.part.content)
                     await send_event(
                         self.websocket,
                         "thinking_delta",
@@ -2642,6 +2676,7 @@ class AgentSession:
                     )
                 elif isinstance(event.delta, ThinkingPartDelta):
                     if event.delta.content_delta:
+                        collected_thinking.append(event.delta.content_delta)
                         await send_event(
                             self.websocket,
                             "thinking_delta",

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/app/services/conversation.py

@@ -524,6 +524,7 @@ class ConversationService:
             conversation_id=conversation_id,
             role=data.role,
             content=data.content,
+            thinking=data.thinking,
             model_name=data.model_name,
             tokens_used=data.tokens_used,
         )

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/backend/pyproject.toml

@@ -332,6 +332,7 @@ ignore = [
 known-first-party = ["app", "cli"]
 
 [tool.ty.terminal]
+# The "warn" rules below are informational — don't let them fail `ty check`.
 error-on-warning = false
 
 [tool.ty.environment]

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/docker-compose.dev.yml

@@ -123,7 +123,7 @@ services:
 {%- if cookiecutter.use_milvus %}
 
   etcd:
-    image: quay.io/coreos/etcd:v3.6.13
+    image: quay.io/coreos/etcd:v3.7.0
     container_name: {{ cookiecutter.project_slug }}_etcd
     command: >
       etcd --advertise-client-urls http://etcd:2379
@@ -162,7 +162,7 @@ services:
     restart: unless-stopped
 
   milvus:
-    image: milvusdb/milvus:v2.6.19
+    image: milvusdb/milvus:v2.6.20
     container_name: {{ cookiecutter.project_slug }}_milvus
     command: ["milvus", "run", "standalone"]
     ports:
@@ -196,7 +196,7 @@ services:
 {%- if cookiecutter.use_qdrant %}
 
   qdrant:
-    image: qdrant/qdrant:v1.18.1
+    image: qdrant/qdrant:v1.18.3
     container_name: {{ cookiecutter.project_slug }}_qdrant
     ports:
       - "6333:6333"

{fastapi_fullstack-0.2.15 → fastapi_fullstack-0.2.16}/template/{{cookiecutter.project_slug}}/docker-compose.prod.yml

@@ -270,7 +270,7 @@ services:
 {%- if cookiecutter.use_milvus %}
 
   etcd:
-    image: quay.io/coreos/etcd:v3.6.13
+    image: quay.io/coreos/etcd:v3.7.0
     container_name: {{ cookiecutter.project_slug }}_etcd
     volumes:
       - etcd_data:/var/lib/etcd
@@ -313,7 +313,7 @@ services:
           memory: 256M
 
   milvus:
-    image: milvusdb/milvus:v2.6.19
+    image: milvusdb/milvus:v2.6.20
     container_name: {{ cookiecutter.project_slug }}_milvus
     command: ["milvus", "run", "standalone"]
     # NO external ports - only accessible within backend-internal network
@@ -348,7 +348,7 @@ services:
 {%- if cookiecutter.use_qdrant %}
 
   qdrant:
-    image: qdrant/qdrant:v1.18.1
+    image: qdrant/qdrant:v1.18.3
     container_name: {{ cookiecutter.project_slug }}_qdrant
     # NO external ports in production - only accessible within backend-internal network
     volumes: