(llm-first): context budget, validation, and eval harness, orchestration general-prompt

tools/eval/fixtures/repair-order/meta.json

@@ -0,0 +1,13 @@
+{
+  "entity": "CategoryResource",
+  "kebab": "category-resource",
+  "resource": "category-resources",
+  "description": "Current FK-heavy entity: UUID PK with references to Part and Employee. Tests reference wiring, autocomplete filters, and protected CRUD routes.",
+  "tests": [
+    "dto-decorator-coverage",
+    "auth-guards",
+    "fk-reference-input",
+    "fk-reference-field",
+    "content-range-header"
+  ]
+}