Fix rerun assessment and candidate extraction

src/repo_scoping/candidate_graph/generator.py

@@ -566,17 +566,41 @@ class CandidateGraphGenerator:
     def _intent_capability_items(self, chunks: list[ContentChunk]) -> list[str]:
         items: list[str] = []
         in_capability_section = False
+        capability_section_level = 0
         for chunk in sorted(chunks, key=lambda item: (item.path, item.start_line)):
             for raw_line in chunk.text.splitlines():
                 line = raw_line.strip()
                 if not line:
                     continue
                 if line.startswith("#"):
-                    heading = line.lstrip("#").strip().lower()
-                    in_capability_section = (
+                    level = len(line) - len(line.lstrip("#"))
+                    heading_text = re.sub(r"\\([._-])", r"\1", line.lstrip("#").strip())
+                    heading = re.sub(
+                        r"^\d+(?:\.\d+)*\.?\s+",
+                        "",
+                        heading_text,
+                    ).lower()
+                    if in_capability_section and level > capability_section_level:
+                        item = re.sub(
+                            r"^\d+(?:\.\d+)*\.?\s+",
+                            "",
+                            heading_text,
+                        )
+                        if item and item.lower() not in {"capabilities", "intended capabilities"}:
+                            items.append(item)
+                        continue
+                    opens_capability_section = (
                         "capabilit" in heading
-                        or heading in {"primary utility", "core utility"}
+                        or heading
+                        in {
+                            "outcomes",
+                            "primary outcomes",
+                            "primary utility",
+                            "core utility",
+                        }
                     )
+                    in_capability_section = opens_capability_section
+                    capability_section_level = level if opens_capability_section else 0
                     continue
                 if not in_capability_section:
                     continue
@@ -594,6 +618,16 @@ class CandidateGraphGenerator:
             return "Make Connectivity Observable Auditable And Controllable"
         if "cli tool" in lowered and "mcp" in lowered:
             return "Expose CLI And MCP Accessible Service"
+        capability_outcomes = {
+            "capability discovery": "Support Capability Discovery",
+            "capability modeling": "Model Capabilities",
+            "capability realisation": "Realize Capabilities",
+            "capability realization": "Realize Capabilities",
+            "capability validation": "Validate Capabilities",
+            "capability evolution": "Evolve Capabilities",
+        }
+        if lowered.strip(" .:-") in capability_outcomes:
+            return capability_outcomes[lowered.strip(" .:-")]
         candidate = re.split(r"\s+-\s+|\s*:\s*|[.!?]\s+", text.strip(), maxsplit=1)[0]
         candidate = candidate.strip(" .:-")
         if not candidate:
@@ -601,6 +635,12 @@ class CandidateGraphGenerator:
         words = candidate.split()
         if words:
             words[0] = self._imperative_verb(words[0])
+            if (
+                len(words) > 1
+                and words[0].lower() in {"analyze", "compare", "detect", "explore", "identify", "interpret"}
+                and words[1].lower().strip(",;:") == "of"
+            ):
+                words.pop(1)
         while words and words[-1].lower().strip(",;:") in {"a", "an", "the", "and", "or", "as", "both"}:
             words.pop()
         return self._title_from_words(words[:10])
@@ -1614,6 +1654,7 @@ class CandidateGraphGenerator:
 
     def _imperative_purpose(self, text: str) -> str:
         cleaned = re.sub(r"\s+", " ", text.strip())
+        cleaned = re.split("\\s+(?:-|\\u2013|\\u2014)\\s+", cleaned, maxsplit=1)[0]
         cleaned = re.split(r"[.!?]\s+", cleaned, maxsplit=1)[0]
         cleaned = re.sub(
             r"(?i)^this\s+repository\s+exists\s+to\s+provide\s+(?:an?\s+)?",
@@ -1635,8 +1676,14 @@ class CandidateGraphGenerator:
             return word
         lower = word.lower().strip(",;:")
         irregular = {
+            "analysis": "analyze",
+            "comparison": "compare",
+            "detection": "detect",
             "does": "do",
+            "exploration": "explore",
             "has": "have",
+            "identification": "identify",
+            "interpretation": "interpret",
             "is": "be",
         }
         if lower in irregular:
@@ -1655,7 +1702,7 @@ class CandidateGraphGenerator:
 
     def _title_from_words(self, words: list[str]) -> str:
         cleaned_words = [
-            re.sub(r"[^A-Za-z0-9_/{}-]", "", word)
+            re.sub(r"[^\w/{}-]", "", word, flags=re.UNICODE)
             for word in words
         ]
         return " ".join(

src/repo_scoping/cli.py

@@ -334,7 +334,7 @@ def dataset_assessment(service: RegistryService) -> dict[str, object]:
     }
     for repository in service.list_repositories():
         runs = service.list_analysis_runs(repository.id)
-        latest_run = next((run for run in reversed(runs) if run.status == "completed"), None)
+        latest_run = next((run for run in runs if run.status == "completed"), None)
         facts = service.list_observed_facts(repository.id, latest_run.id) if latest_run else []
         chunks = service.list_content_chunks(repository.id, latest_run.id) if latest_run else []
         candidate_counts = {

src/repo_scoping/core/service.py

@@ -1581,6 +1581,7 @@ class RegistryService:
                     "kind": kind,
                     "layer": self._dependency_layer(kind),
                     "label": detail.get("label")
+                    or detail.get("name")
                     or self._dependency_node_label(repository_id, kind, key, item_id),
                     "reviewState": detail.get("reviewState", "accepted"),
                     "name": detail.get("name")
@@ -2724,7 +2725,7 @@ class RegistryService:
             for run in self.store.list_analysis_runs(repository_id)
             if run.status == "completed"
         ]
-        return completed[-1] if completed else None
+        return completed[0] if completed else None
 
     def _candidate_graph_or_none(
         self,