feat: Add fast startup fetch and background caching

- Add  for quick startup using first N feeds
- Add background thread for full fetch and caching
- Update  to use fast fetch
- Update docs and skills

.opencode/skills/mainline-architecture/SKILL.md

@@ -29,17 +29,28 @@ class Stage(ABC):
         return set()
     
     @property
-    def dependencies(self) -> list[str]:
-        """What this stage needs (e.g., ['source'])"""
-        return []
+    def dependencies(self) -> set[str]:
+        """What this stage needs (e.g., {'source'})"""
+        return set()
 ```
 
 ### Capability-Based Dependencies
 
 The Pipeline resolves dependencies using **prefix matching**:
 - `"source"` matches `"source.headlines"`, `"source.poetry"`, etc.
+- `"camera.state"` matches the camera state capability
 - This allows flexible composition without hardcoding specific stage names
 
+### Minimum Capabilities
+
+The pipeline requires these minimum capabilities to function:
+- `"source"` - Data source capability
+- `"render.output"` - Rendered content capability
+- `"display.output"` - Display output capability
+- `"camera.state"` - Camera state for viewport filtering
+
+These are automatically injected if missing (auto-injection).
+
 ### DataType Enum
 
 PureData-style data types for inlet/outlet validation:
@@ -76,3 +87,11 @@ Canvas tracks dirty regions automatically when content is written via `put_regio
 - Use adapters (engine/pipeline/adapters.py) to wrap existing components as stages
 - Set `optional=True` for stages that can fail gracefully
 - Use `stage_type` and `render_order` for execution ordering
+- Clock stages update state independently of data flow
+
+## Sources
+
+- engine/pipeline/core.py - Stage base class
+- engine/pipeline/controller.py - Pipeline implementation
+- engine/pipeline/adapters/ - Stage adapters
+- docs/PIPELINE.md - Pipeline documentation