diff --git a/CHANGELOG.md b/CHANGELOG.md
index ef83812d..64d1e776 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -129,6 +129,9 @@ All notable changes to SkillOpt are documented here. This project adheres to
   value is rejected with a clear error.
 
 ### Fixed
+- `insert_after` with a target that ends in a newline now inserts right
+  after that line instead of after the next one, which could file the new
+  text under the following heading or inside a protected region (#407).
 - Close four ways invalid evidence could reach the SkillOpt-Sleep gate: recall
   no longer brings a content twin of tonight's held-out task into training,
   a measured tool route no longer credits a self-reported `TOOL_CALL:` marker,
diff --git a/skillopt/optimizer/skill.py b/skillopt/optimizer/skill.py
index 69229f94..26ad5625 100644
--- a/skillopt/optimizer/skill.py
+++ b/skillopt/optimizer/skill.py
@@ -119,8 +119,13 @@ def _apply_edit_with_report(skill: str, edit: EditType | dict) -> tuple[str, dic
             report["status"] = "applied_insert_after_fallback_append"
             return skill.rstrip() + "\n\n" + content + "\n", report
         idx = skill.index(target) + len(target)
-        newline = skill.find("\n", idx)
-        insert_at = newline + 1 if newline != -1 else len(skill)
+        if target.endswith("\n"):
+            # The target already ends its line: insert right here, not after
+            # the line that follows it.
+            insert_at = idx
+        else:
+            newline = skill.find("\n", idx)
+            insert_at = newline + 1 if newline != -1 else len(skill)
         report["status"] = "applied_insert_after"
         return skill[:insert_at] + "\n" + content + "\n" + skill[insert_at:], report
 
diff --git a/tests/test_insert_after_newline_target.py b/tests/test_insert_after_newline_target.py
new file mode 100644
index 00000000..a5503ed0
--- /dev/null
+++ b/tests/test_insert_after_newline_target.py
@@ -0,0 +1,44 @@
+"""Regression: ``insert_after`` with a target that ends in a newline.
+
+The insertion point is the end of the target's line. When the target already
+ends with ``\\n``, the search for the next newline found the end of the
+*following* line, so the content landed one line too late: under the next
+heading, or inside a protected region whose start marker follows directly.
+"""
+from __future__ import annotations
+
+from skillopt.optimizer.skill import apply_patch_with_report
+
+
+def _insert_after(skill: str, target: str) -> tuple[str, str]:
+    new, rep = apply_patch_with_report(
+        skill,
+        {"edits": [{"op": "insert_after", "target": target, "content": "- Cite the source."}]},
+    )
+    return new, rep[0]["status"]
+
+
+def test_target_ending_in_newline_inserts_right_after_it() -> None:
+    skill = "# QA Skill\n\n## Rules\n## Output format\n- One line.\n"
+    new, status = _insert_after(skill, "## Rules\n")
+    assert status == "applied_insert_after"
+    assert new == "# QA Skill\n\n## Rules\n\n- Cite the source.\n## Output format\n- One line.\n"
+
+
+def test_target_ending_in_newline_matches_target_without_it() -> None:
+    skill = "# QA Skill\n\n## Rules\n## Output format\n- One line.\n"
+    with_newline, _ = _insert_after(skill, "## Rules\n")
+    without_newline, _ = _insert_after(skill, "## Rules")
+    assert with_newline == without_newline
+
+
+def test_insert_does_not_land_inside_following_protected_region() -> None:
+    skill = (
+        "# QA Skill\n- Answer briefly.\n"
+        "<!-- SLOW_UPDATE_START -->\nPrefer exact spans.\n<!-- SLOW_UPDATE_END -->\n"
+    )
+    new, status = _insert_after(skill, "- Answer briefly.\n")
+    assert status == "applied_insert_after"
+    start = new.index("<!-- SLOW_UPDATE_START -->")
+    assert new.index("- Cite the source.") < start
+    assert new[start:] == skill[skill.index("<!-- SLOW_UPDATE_START -->"):]