Cross repo isolation

src/repo_registry/llm_extraction/extractor.py

@@ -75,12 +75,20 @@ class LLMCandidateExtractor:
         chunk_text = "\n\n".join(
             (
                 f"Source: {chunk.path}:{chunk.start_line}-{chunk.end_line} "
-                f"({chunk.kind})\n{chunk.text}"
+                f"({chunk.kind}; source_role={self._source_role(chunk)})\n{chunk.text}"
             )
-            for chunk in chunks[:12]
+            for chunk in self._prompt_chunks(chunks)
         )
         return (
             "Extract a conservative, source-linked repository ability map.\n"
+            "Use original repository utility only: capabilities the repository "
+            "owns, intentionally exposes as a facade, or implements as an adapter.\n"
+            "Prefer source_role=intent_summary, product_documentation, "
+            "implementation_source, and test_evidence. Do not use SCOPE.md or "
+            "source_role=derived_scope as primary evidence; it is a derived prior "
+            "registry view and may be stale. Ignore agent guidance, CI/tooling, "
+            "dependency-only, and mention-only context unless owned product "
+            "evidence supports the same claim.\n"
             "Return strict JSON only with this shape:\n"
             "{\n"
             '  "abilities": [\n'
@@ -108,6 +116,46 @@ class LLMCandidateExtractor:
             f"{chunk_text}\n"
         )
 
+    def _prompt_chunks(self, chunks: list[ContentChunk]) -> list[ContentChunk]:
+        promptable = [
+            chunk
+            for chunk in chunks
+            if self._source_role(chunk) not in {"agent_guidance", "derived_scope"}
+        ]
+        return sorted(
+            promptable,
+            key=lambda chunk: (
+                self._source_role_priority(self._source_role(chunk)),
+                chunk.path,
+                chunk.start_line,
+            ),
+        )[:12]
+
+    def _source_role(self, chunk: ContentChunk) -> str:
+        role = chunk.metadata.get("source_role")
+        if isinstance(role, str) and role:
+            return role
+        path = chunk.path.lower()
+        if path.endswith("intent.md"):
+            return "intent_summary"
+        if path.endswith("scope.md"):
+            return "derived_scope"
+        if path.endswith(("agents.md", "claude.md")) or "/.claude/" in path:
+            return "agent_guidance"
+        return ""
+
+    def _source_role_priority(self, source_role: str) -> int:
+        priorities = {
+            "intent_summary": 0,
+            "product_documentation": 1,
+            "implementation_source": 2,
+            "test_evidence": 3,
+            "configuration": 4,
+            "dependency_declaration": 5,
+            "ci_tooling": 6,
+        }
+        return priorities.get(source_role, 7)
+
     def parse_response(self, content: str) -> list[ExtractedAbility]:
         try:
             payload = json.loads(self._json_text(content))