feat: add compose deliverable, scenario definitions, eval stub, and CI

eval/run_eval.py

@@ -0,0 +1,23 @@
+"""Eval harness entry point.
+
+Fixture mode (default, key-free): byte-snapshots of deterministic pipeline
+stages against recorded cassettes. Live mode: property assertions only —
+LLM output is constrained, not pinned.
+"""
+
+from pathlib import Path
+
+import yaml
+
+SCENARIOS_FILE = Path(__file__).parent / "scenarios.yaml"
+
+
+def load_scenarios() -> dict:
+    """Load the shared scenario definitions."""
+    with SCENARIOS_FILE.open() as handle:
+        return yaml.safe_load(handle)
+
+
+if __name__ == "__main__":
+    scenarios = load_scenarios()
+    print(f"{len(scenarios['scenarios'])} scenarios, {len(scenarios['refinements'])} refinements")

eval/scenarios.yaml

@@ -0,0 +1,45 @@
+# One source of truth for three consumers: golden eval queries, demo fixture
+# keys, and UI suggestion chips. Every chip therefore has a fixture, and every
+# fixture has a golden query.
+
+scenarios:
+  - key: focus-coding
+    chip: "Focus while coding"
+    query: "something calm for while I'm programming, but not boring"
+    facets: { activity: coding, familiarity: mixed }
+
+  - key: workout-energy
+    chip: "Energy for the gym"
+    query: "high-energy music for a heavy workout"
+    facets: { activity: workout, mood: energetic }
+
+  - key: dutch-chill
+    chip: "Dutch-language & relaxed"
+    query: "relaxed Dutch-language music for lounging on the couch"
+    facets: { language: nl, mood: relaxed }
+
+  - key: nineties-nostalgia
+    chip: "90s nostalgia"
+    query: "take me back to the nineties"
+    facets: { era: "1990s" }
+
+  - key: discover-new
+    chip: "Surprise me with something new"
+    query: "something I don't know yet but will probably like"
+    facets: { familiarity: new } # asserts known-track exclusion
+
+  - key: rainy-sunday
+    chip: "Rainy Sunday"
+    query: "melancholic but warm, for a rainy Sunday morning"
+    facets: { mood: melancholic }
+
+  - key: dinner-background
+    chip: "Background for dinner"
+    query: "something jazzy for during dinner, not too present"
+    facets: { mood: understated }
+
+# Multi-turn refinement goldens: reuse the grounded pool, no re-grounding.
+refinements:
+  - key: focus-coding-refine
+    after: focus-coding
+    query: "a bit more electronic, and drop number 3"