Fix rerun assessment and candidate extraction

workplans/RREG-WP-0018-agentic-hierarchy-and-intent-scope-review.md

@@ -235,8 +235,37 @@ dataset:
   scope-vs-intent separation; deterministic gates can require review but do not
   accept registry truth.
 
-The latest local assessment command currently sees nine repositories because
-`vantage-point` has been added. It still reports old sparse Railiance candidate
-counts because those stored analysis runs predate this implementation. T07 stays
-open until the affected repositories are rerun and compared against the sparse
-baseline.
+The latest local assessment command initially saw nine repositories because
+`vantage-point` had been added. It still reported old sparse Railiance candidate
+counts because those stored analysis runs predated this implementation. T07
+stays open until the affected repositories are rerun and compared against the
+sparse baseline.
+
+## Rerun Review 2026-05-16
+
+The local dataset now contains ten repositories and several post-implementation
+reruns. A review found that `assess-dataset` and the dependency graph fallback
+were incorrectly selecting the oldest completed analysis run because
+`list_analysis_runs` is sorted newest-first. That has been corrected.
+
+Corrected assessment results:
+- Dataset total: `10` repos, `430` facts, candidate hierarchy
+  `10/26/36/44`, graph `210/387`.
+- Improved: `railiance-cluster` now has `3` capabilities / `3` features;
+  `railiance-platform` has `3` / `3`; `railiance-enablement` has `2` / `2`;
+  `ops-warden` has repo-specific scope naming and `1` / `2`;
+  `vergabe-teilnahme` has `1` / `4`.
+- Still sparse because they were not rerun after the implementation:
+  `railiance-infra` and `railiance-apps`. Read-only generator preview shows
+  they would now produce `3` and `1` scope-derived capabilities respectively.
+- New sparse repo: `helix-forge`. Its `INTENT.md` uses numbered/escaped
+  `Primary outcomes` sections rather than bullet-based intended capabilities;
+  generator support was added for this shape and preview now yields five
+  outcome-derived capabilities.
+- Naming polish added for reviewability: preserve non-ASCII letters, normalize
+  nominalized capability names such as `Analysis of...` to `Analyze...`, and
+  trim explanatory dash clauses from scope one-liners.
+
+T07 remains in progress until `railiance-infra`, `railiance-apps`, and
+`helix-forge` are rerun and the corrected assessment report is captured as the
+comparison artifact.