Feature/theme suggestions

.env.test

@@ -5,3 +5,5 @@ OLLAMA_URL=http://ollama:11434
 REDDIT_CLIENT_ID=client
 REDDIT_CLIENT_SECRET=secret
 REDDIT_USER_AGENT=newsletter-maker/test
+CELERY_BROKER_URL=memory://
+CELERY_RESULT_BACKEND=cache+memory://

.github/copilot-instructions.md

@@ -45,6 +45,12 @@ You are working in Newsletter Maker, a Django + DRF + Celery + Qdrant backend wi
 - TypeScript and React code should use JSDoc for exported utilities, hooks, route handlers, and non-trivial components when behavior is not obvious from the type signature alone.
 - If architecture or workflow behavior changes, update the most relevant docs in `docs/`, especially `docs/DEVELOPER_GUIDE.md`, `docs/IMPLEMENTATION_OVERVIEW.md`, `docs/MODELS.md`, `docs/RELEVANCE_SCORING.md`, or `docs/LOGGING.md`.
 
+## Prompt Skill Conventions
+
+- Application prompt skills live under `skills/<skill_name>/SKILL.md` and are loaded by `core/llm.py` using the folder name rather than the frontmatter `name`.
+- When adding or editing one of these prompt skills, always include a short frontmatter `description` so VS Code does not report incomplete skill metadata.
+- If a frontmatter `name` is present, prefer lowercase letters, numbers, and hyphens there to satisfy the Copilot markdown validator, even when the runtime skill key elsewhere in the app still uses underscores.
+
 ## Testing And Validation
 
 - Backend tests use `pytest`.

.github/instructions/prompt-skills.instructions.md

@@ -0,0 +1,15 @@
+---
+name: "Prompt Skill Markdown Guidelines"
+description: "Use when editing application prompt skills in skills/**/SKILL.md. Covers custom frontmatter, runtime lookup behavior, and keeping VS Code's skill validator quiet."
+applyTo:
+  - "skills/**/SKILL.md"
+---
+
+# Prompt Skill Markdown Guidelines
+
+- Files under `skills/**/SKILL.md` are application prompt specs loaded by `core/llm.py`, not Copilot repo skills under `.github/skills/`.
+- The runtime skill key is the folder name, so keep folder names and code constants aligned with the app's expected key.
+- Always include a short frontmatter `description` to avoid VS Code skill-schema warnings.
+- If a frontmatter `name` is present, prefer lowercase letters, numbers, and hyphens so the Copilot validator stays quiet, even when the runtime key elsewhere in the app uses underscores.
+- Preserve the repo's `input` and `output` frontmatter fields because `core/llm.py` reads them to build prompts.
+- Keep the body concise and instruction-focused. Return-shape requirements should stay explicit in the markdown body.

.vscode/settings.json

@@ -1,28 +1,45 @@
 {
   "cSpell.words": [
+    "abangser",
     "ASGI",
     "botocore",
+    "bsky",
     "buildx",
     "cbranch",
     "cfgv",
     "cstat",
+    "dabde",
     "dateutil",
     "djlint",
     "dnspython",
+    "doseq",
+    "exif",
     "falsey",
+    "Farcic",
     "FAVICONS",
+    "favourites",
+    "fbclid",
     "Feedly",
     "Fraunces",
+    "gclid",
     "gunicorn",
+    "Hashimoto",
     "healthz",
     "HNSW",
     "httpx",
+    "Hykes",
     "isready",
     "jsbeautifier",
+    "LANCZOS",
     "libipld",
     "libpq",
     "librt",
+    "lnkd",
     "mday",
+    "membe",
+    "mipsytipsy",
+    "mitchellh",
+    "newsle",
     "nodeenv",
     "noinput",
     "nomic",
@@ -39,14 +56,24 @@
     "readyz",
     "recomputations",
     "Referer",
+    "repost",
+    "reposts",
+    "rollup",
     "scaffolder",
     "simplejwt",
+    "skillr",
+    "solomonstre",
     "svix",
+    "topicv",
     "Unparseable",
     "unstub",
     "upserted",
+    "upserts",
     "upvote",
     "uritemplate",
+    "Vercel",
+    "vfarcic",
+    "Viktor",
     "xrpc",
     "xxhash"
   ]

content/migrations/0002_content_newsletter_promotion.py

@@ -0,0 +1,41 @@
+import django.db.models.deletion
+from django.conf import settings
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        migrations.swappable_dependency(settings.AUTH_USER_MODEL),
+        ("content", "0001_initial"),
+        ("trends", "0003_theme_suggestion"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="content",
+            name="newsletter_promotion_at",
+            field=models.DateTimeField(blank=True, null=True),
+        ),
+        migrations.AddField(
+            model_name="content",
+            name="newsletter_promotion_by",
+            field=models.ForeignKey(
+                blank=True,
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="newsletter_promoted_content",
+                to=settings.AUTH_USER_MODEL,
+            ),
+        ),
+        migrations.AddField(
+            model_name="content",
+            name="newsletter_promotion_theme",
+            field=models.ForeignKey(
+                blank=True,
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="promoted_contents",
+                to="trends.themesuggestion",
+            ),
+        ),
+    ]

content/models.py

@@ -47,6 +47,21 @@ class Content(models.Model):
     duplicate_signal_count = models.IntegerField(default=0)
     is_reference = models.BooleanField(default=False)
     is_active = models.BooleanField(default=True)
+    newsletter_promotion_at = models.DateTimeField(null=True, blank=True)
+    newsletter_promotion_by = models.ForeignKey(
+        settings.AUTH_USER_MODEL,
+        null=True,
+        blank=True,
+        on_delete=models.SET_NULL,
+        related_name="newsletter_promoted_content",
+    )
+    newsletter_promotion_theme = models.ForeignKey(
+        "trends.ThemeSuggestion",
+        null=True,
+        blank=True,
+        on_delete=models.SET_NULL,
+        related_name="promoted_contents",
+    )
 
     class Meta:
         ordering = ["-published_date"]

content/serializers.py

@@ -32,6 +32,9 @@ class Meta:
             "duplicate_signal_count",
             "is_reference",
             "is_active",
+            "newsletter_promotion_at",
+            "newsletter_promotion_by",
+            "newsletter_promotion_theme",
         ]
         read_only_fields = [
             "id",
@@ -42,6 +45,9 @@ class Meta:
             "embedding_id",
             "duplicate_of",
             "duplicate_signal_count",
+            "newsletter_promotion_at",
+            "newsletter_promotion_by",
+            "newsletter_promotion_theme",
         ]
 
     def validate(self, attrs):

core/api.py

@@ -6,7 +6,7 @@
 """
 
 import logging
-from typing import Any
+from typing import Any, cast
 
 from drf_spectacular.utils import (
     OpenApiExample,
@@ -589,6 +589,16 @@ class ProjectOwnedQuerysetMixin:
 
     queryset: Any = None
 
+    def _kwargs(self) -> dict[str, Any]:
+        """Return the DRF route kwargs for typed nested-project lookups."""
+
+        return cast(dict[str, Any], getattr(self, "kwargs"))
+
+    def _request(self) -> Any:
+        """Return the DRF request object for typed access checks."""
+
+        return getattr(self, "request")
+
     def get_project(self):
         """Return the project referenced by ``project_id`` after access checks.
 
@@ -597,13 +607,15 @@ def get_project(self):
             NotFound: If the project does not exist or the user lacks access.
         """
 
-        project_id = self.kwargs.get("project_id")
+        project_id = self._kwargs().get("project_id")
         if project_id is None:
             raise AssertionError(
                 "project_id must be present in nested project-scoped routes"
             )
         try:
-            return get_visible_projects_queryset(self.request.user).get(pk=project_id)
+            return get_visible_projects_queryset(self._request().user).get(
+                pk=project_id
+            )
         except Project.DoesNotExist as exc:
             raise NotFound("Project not found.") from exc
 
@@ -618,7 +630,7 @@ def get_queryset(self):
     def get_serializer_context(self):
         """Inject the resolved project into serializer context."""
 
-        context = super().get_serializer_context()
+        context = cast(dict[str, Any], cast(Any, super()).get_serializer_context())
         context["project"] = self.get_project()
         return context
 

core/embeddings.py

@@ -15,6 +15,7 @@
 
 import httpx
 from django.conf import settings as django_settings
+from django.db.models import Model
 from django.utils.dateparse import parse_datetime
 from qdrant_client import QdrantClient
 from qdrant_client.models import (
@@ -34,6 +35,15 @@
 settings = cast(CoreSettings, django_settings)
 
 
+def _require_pk(instance: Model) -> int:
+    """Return a saved model primary key for typed Qdrant payload construction."""
+
+    instance_pk = instance.pk
+    if instance_pk is None:
+        raise ValueError(f"{instance.__class__.__name__} must be saved first.")
+    return int(instance_pk)
+
+
 def get_sentence_transformer_class():
     """Lazily import and cache the sentence-transformer class.
 
@@ -90,7 +100,10 @@ def embed_text(self, text: str) -> list[float]:
     def get_embedding_dimension(self) -> int:
         """Return the model's native embedding dimension without probing text."""
 
-        return int(self.model.get_sentence_embedding_dimension())
+        dimension = self.model.get_sentence_embedding_dimension()
+        if dimension is None:
+            raise RuntimeError("Embedding model did not report a vector dimension.")
+        return int(dimension)
 
 
 class OllamaEmbeddingProvider(EmbeddingProvider):
@@ -222,18 +235,20 @@ def upsert_content_embedding(content: Content) -> str:
     """
 
     client = get_qdrant_client()
-    ensure_project_collection(content.project_id)
+    project_id = _require_pk(content.project)
+    content_id = _require_pk(content)
+    ensure_project_collection(project_id)
     embedding_id = content.embedding_id or str(uuid4())
     vector = embed_text(build_content_embedding_text(content))
     client.upsert(
-        collection_name=collection_name_for_project(content.project_id),
+        collection_name=collection_name_for_project(project_id),
         points=[
             PointStruct(
                 id=embedding_id,
                 vector=vector,
                 payload={
-                    "content_id": content.id,
-                    "project_id": content.project_id,
+                    "content_id": content_id,
+                    "project_id": project_id,
                     "url": content.url,
                     "title": content.title,
                     "published_date": serialize_published_date(content.published_date),
@@ -254,18 +269,20 @@ def upsert_entity_embedding(entity: Entity) -> str:
     """Write or update an entity embedding in the project's entity collection."""
 
     client = get_qdrant_client()
-    ensure_project_entity_collection(entity.project_id)
+    project_id = _require_pk(entity.project)
+    entity_id = _require_pk(entity)
+    ensure_project_entity_collection(project_id)
     vector = embed_text(build_entity_embedding_text(entity))
-    embedding_id = f"entity-{entity.id}"
+    embedding_id = f"entity-{entity_id}"
     client.upsert(
-        collection_name=entity_collection_name_for_project(entity.project_id),
+        collection_name=entity_collection_name_for_project(project_id),
         points=[
             PointStruct(
                 id=embedding_id,
                 vector=vector,
                 payload={
-                    "entity_id": entity.id,
-                    "project_id": entity.project_id,
+                    "entity_id": entity_id,
+                    "project_id": project_id,
                     "name": entity.name,
                     "type": entity.type,
                 },
@@ -343,11 +360,11 @@ def search_similar_content(
     """Find content similar to an existing content row within the same project."""
 
     return search_similar(
-        content.project_id,
+        _require_pk(content.project),
         embed_text(build_content_embedding_text(content)),
         limit=limit,
         is_reference=is_reference,
-        exclude_content_id=content.id,
+        exclude_content_id=_require_pk(content),
     )
 
 
@@ -370,9 +387,10 @@ def search_similar_entities(
 def search_similar_entities_for_content(content: Content, limit: int = 8):
     """Find tracked entities whose embeddings are close to a content item."""
 
-    sync_project_entity_embeddings(content.project_id)
+    project_id = _require_pk(content.project)
+    sync_project_entity_embeddings(project_id)
     return search_similar_entities(
-        content.project_id,
+        project_id,
         embed_text(build_content_embedding_text(content)),
         limit=limit,
     )