Escape lineage metadata and surface actionable diagnostics

CHANGELOG.md

@@ -15,6 +15,15 @@ shadow knowledge bases for any codebase.
   repeated guidance and examples in the core, Dream, Init, Meditate, Update,
   and Viewer skills while preserving data formats, policy limits, and safety gates.
 
+### Fixed
+- **Lineage error feedback** — required input/output failures now name the
+  affected path and repair action on stderr. Recoverable metadata omissions
+  and malformed rows produce visible warnings instead of silent fallbacks;
+  invalid explicit shadow paths no longer select a different shadow.
+- **Lineage HTML escaping** — render experiment names (including branch fallbacks)
+  and manifest test counts as literal text in timeline and compact tree views,
+  without changing stored metadata or empty/zero count display behavior (#37).
+
 ---
 
 ## 2026-09-22

skills/shadow-frog-viewer/SKILL.md

@@ -83,6 +83,15 @@ python3 .github/skills/shadow-frog-viewer/dream-lineage.py --shadow-dir /path/to
 
 The HTML groups compounding chains and fresh experiments, includes a full
 lineage tree, and supports expanding each experiment's report.
+Experiment names and test-count metadata render as literal text, not HTML;
+escaping happens at rendering time without changing the stored metadata.
+
+Pass stderr back to the agent even when generation exits 0. Required directory,
+index, or output failures emit `ERROR` and exit 1; malformed optional metadata
+or omitted rows emit `WARNING` with the affected path/field and produce a
+partial view. Repair the indicated input or permissions and rerun the same
+command. Missing optional artifacts and safely escaped text are not errors.
+The helper does not invoke a model or retry automatically.
 
 ## Fallback: Shell One-Liners