diff --git a/src/specify_cli/integrations/claude/__init__.py b/src/specify_cli/integrations/claude/__init__.py
index 31972c4b0..34ed8411e 100644
--- a/src/specify_cli/integrations/claude/__init__.py
+++ b/src/specify_cli/integrations/claude/__init__.py
@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import re
 from pathlib import Path
 from typing import Any
 
@@ -24,6 +25,61 @@
     "taskstoissues": "Optional filter or label for GitHub issues",
 }
 
+# Begin/end markers used to fence question-rendering blocks in command
+# templates.  The post-processor replaces the content between these markers
+# with Claude Code-native AskUserQuestion instructions.
+_QUESTION_FENCE_BEGIN = "<!-- speckit:question-render:begin -->"
+_QUESTION_FENCE_END = "<!-- speckit:question-render:end -->"
+
+# Replacement block for /clarify.  Maps the `Option | Description` table
+# schema to AskUserQuestion's `{label, description}` shape.
+# The recommended option is placed first with a "Recommended — <reasoning>"
+# description prefix.
+_CLARIFY_ASK_USER = """\
+    - For multiple‑choice questions:
+       - **Analyze all options** and determine the **most suitable option** based on:
+          - Best practices for the project type
+          - Common patterns in similar implementations
+          - Risk reduction (security, performance, maintainability)
+          - Alignment with any explicit project goals or constraints visible in the spec
+       - Use the `AskUserQuestion` tool to present the question as a native structured picker:
+         - `question`: the clarification question text, prefixed with your recommendation:
+           "Recommended: Option [X] — <1-2 sentence reasoning>\\n\\n<question text>"
+         - `options[]`: an array of `{label, description}` objects. Place the **recommended option first** and prefix its `description` with `Recommended — <reasoning>.`
+           Build each option as: `{label: "<A|B|C|...>", description: "<option description>"}`.
+         - Append a final option: `{label: "Short", description: "Provide my own short answer (≤5 words)"}` to preserve the free-form escape hatch.
+         - `multiSelect`: `false`
+       - If the user selects the "Short" option, ask a follow-up free-text question constrained to ≤5 words.
+    - For short‑answer style (no meaningful discrete options):
+       - Determine your **suggested answer** based on best practices and context.
+       - Use the `AskUserQuestion` tool:
+         - `question`: "Suggested: <your proposed answer> — <brief reasoning>\\n\\n<question text>\\nFormat: Short answer (≤5 words)."
+         - `options[]`: `[{label: "Accept suggestion", description: "Use the suggested answer above"}, {label: "Custom", description: "Provide my own short answer (≤5 words)"}]`
+         - `multiSelect`: `false`
+       - If the user selects "Custom", ask a follow-up free-text question constrained to ≤5 words."""
+
+# Replacement block for /checklist.  Maps the `Option | Candidate | Why It
+# Matters` table schema to AskUserQuestion's `{label, description}` shape:
+#   - "Candidate" becomes the option `label`
+#   - "Why It Matters" becomes the option `description`
+_CHECKLIST_ASK_USER = """\
+   Question formatting rules:
+   - If presenting options, use the `AskUserQuestion` tool to present a native structured picker:
+     - `question`: the clarification question text
+     - `options[]`: an array of `{label, description}` objects. For each candidate option: `{label: "<Candidate value>", description: "<Why It Matters value>"}`.
+     - Append a final option: `{label: "Custom", description: "Provide my own short answer (≤5 words)"}` to preserve the free-form escape hatch.
+     - `multiSelect`: `false`
+   - If the user selects the "Custom" option, ask a follow-up free-text question.
+   - Omit the picker if a free-form answer is clearer (use `AskUserQuestion` with only a `question` and no `options[]` in that case).
+   - Never ask the user to restate what they already said.
+   - Avoid speculative categories (no hallucination). If uncertain, ask explicitly: "Confirm whether X belongs in scope.\""""
+
+# Map of skill stem → replacement content for the question-rendering fence.
+_QUESTION_RENDER_REPLACEMENTS: dict[str, str] = {
+    "clarify": _CLARIFY_ASK_USER,
+    "checklist": _CHECKLIST_ASK_USER,
+}
+
 
 class ClaudeIntegration(SkillsIntegration):
     """Integration for Claude Code skills."""
@@ -44,6 +100,21 @@ class ClaudeIntegration(SkillsIntegration):
     }
     context_file = "CLAUDE.md"
 
+    @staticmethod
+    def replace_question_render_block(content: str, replacement: str) -> str:
+        """Replace the fenced question-rendering block with *replacement*.
+
+        Looks for ``<!-- speckit:question-render:begin -->`` …
+        ``<!-- speckit:question-render:end -->`` and swaps the entire fence
+        (markers included) with *replacement*.  Returns *content* unchanged
+        when no fence is found.
+        """
+        pattern = re.compile(
+            re.escape(_QUESTION_FENCE_BEGIN) + r".*?" + re.escape(_QUESTION_FENCE_END),
+            re.DOTALL,
+        )
+        return pattern.sub(replacement, content, count=1)
+
     @staticmethod
     def inject_argument_hint(content: str, hint: str) -> str:
         """Insert ``argument-hint`` after the first ``description:`` in YAML frontmatter.
@@ -188,6 +259,11 @@ def setup(
             if hint:
                 updated = self.inject_argument_hint(updated, hint)
 
+            # Replace question-rendering fences with AskUserQuestion instructions
+            replacement = _QUESTION_RENDER_REPLACEMENTS.get(stem, "")
+            if replacement:
+                updated = self.replace_question_render_block(updated, replacement)
+
             if updated != content:
                 path.write_bytes(updated.encode("utf-8"))
                 self.record_file_in_manifest(path, project_root, manifest)
diff --git a/templates/commands/checklist.md b/templates/commands/checklist.md
index 533046566..53607f0e8 100644
--- a/templates/commands/checklist.md
+++ b/templates/commands/checklist.md
@@ -93,11 +93,13 @@ You **MUST** consider the user input before proceeding (if not empty).
       - Boundary exclusion (e.g., "Should we explicitly exclude performance tuning items this round?")
       - Scenario class gap (e.g., "No recovery flows detected—are rollback / partial failure paths in scope?")
 
+<!-- speckit:question-render:begin -->
    Question formatting rules:
    - If presenting options, generate a compact table with columns: Option | Candidate | Why It Matters
    - Limit to A–E options maximum; omit table if a free-form answer is clearer
    - Never ask the user to restate what they already said
    - Avoid speculative categories (no hallucination). If uncertain, ask explicitly: "Confirm whether X belongs in scope."
+<!-- speckit:question-render:end -->
 
    Defaults when interaction impossible:
    - Depth: Standard
diff --git a/templates/commands/clarify.md b/templates/commands/clarify.md
index d6d6bbe91..9e369a342 100644
--- a/templates/commands/clarify.md
+++ b/templates/commands/clarify.md
@@ -135,6 +135,7 @@ Execution steps:
 
 4. Sequential questioning loop (interactive):
     - Present EXACTLY ONE question at a time.
+<!-- speckit:question-render:begin -->
     - For multiple‑choice questions:
        - **Analyze all options** and determine the **most suitable option** based on:
           - Best practices for the project type
@@ -157,6 +158,7 @@ Execution steps:
        - Provide your **suggested answer** based on best practices and context.
        - Format as: `**Suggested:** <your proposed answer> - <brief reasoning>`
        - Then output: `Format: Short answer (<=5 words). You can accept the suggestion by saying "yes" or "suggested", or provide your own answer.`
+<!-- speckit:question-render:end -->
     - After the user answers:
        - If the user replies with "yes", "recommended", or "suggested", use your previously stated recommendation/suggestion as the answer.
        - Otherwise, validate the answer maps to one option or fits the <=5 word constraint.
diff --git a/tests/integrations/test_integration_claude.py b/tests/integrations/test_integration_claude.py
index 7fd69df17..6aa50efb3 100644
--- a/tests/integrations/test_integration_claude.py
+++ b/tests/integrations/test_integration_claude.py
@@ -400,3 +400,167 @@ def test_inject_argument_hint_skips_if_already_present(self):
         lines = result.splitlines()
         hint_count = sum(1 for ln in lines if ln.startswith("argument-hint:"))
         assert hint_count == 1
+
+
+class TestQuestionRenderReplacement:
+    """Verify that question-render fences are replaced with AskUserQuestion for Claude skills."""
+
+    def test_clarify_skill_uses_ask_user_question(self, tmp_path):
+        """speckit-clarify/SKILL.md must contain AskUserQuestion instructions,
+        not the Option | Description table."""
+        i = get_integration("claude")
+        m = IntegrationManifest("claude", tmp_path)
+        i.setup(tmp_path, m, script_type="sh")
+
+        skill_file = tmp_path / ".claude" / "skills" / "speckit-clarify" / "SKILL.md"
+        assert skill_file.exists()
+        content = skill_file.read_text(encoding="utf-8")
+
+        # Must contain AskUserQuestion instructions
+        assert "AskUserQuestion" in content
+        # Must NOT contain the original Markdown table header
+        assert "| Option | Description |" not in content
+        # Must NOT contain the fence markers
+        assert "speckit:question-render:begin" not in content
+        assert "speckit:question-render:end" not in content
+        # Must contain the free-form escape hatch option
+        assert "Provide my own short answer" in content
+
+    def test_checklist_skill_uses_ask_user_question(self, tmp_path):
+        """speckit-checklist/SKILL.md must contain AskUserQuestion instructions,
+        not the Option | Candidate | Why It Matters table."""
+        i = get_integration("claude")
+        m = IntegrationManifest("claude", tmp_path)
+        i.setup(tmp_path, m, script_type="sh")
+
+        skill_file = tmp_path / ".claude" / "skills" / "speckit-checklist" / "SKILL.md"
+        assert skill_file.exists()
+        content = skill_file.read_text(encoding="utf-8")
+
+        # Must contain AskUserQuestion instructions
+        assert "AskUserQuestion" in content
+        # Must NOT contain the original table schema
+        assert "| Candidate | Why It Matters" not in content
+        # Must NOT contain the fence markers
+        assert "speckit:question-render:begin" not in content
+        assert "speckit:question-render:end" not in content
+
+    def test_clarify_recommended_option_first(self, tmp_path):
+        """The Claude clarify skill must instruct placing the recommended option first."""
+        i = get_integration("claude")
+        m = IntegrationManifest("claude", tmp_path)
+        i.setup(tmp_path, m, script_type="sh")
+
+        skill_file = tmp_path / ".claude" / "skills" / "speckit-clarify" / "SKILL.md"
+        content = skill_file.read_text(encoding="utf-8")
+
+        assert "recommended option first" in content.lower()
+        assert "Recommended —" in content
+
+    def test_non_question_skills_unchanged(self, tmp_path):
+        """Skills other than clarify/checklist must NOT contain AskUserQuestion."""
+        i = get_integration("claude")
+        m = IntegrationManifest("claude", tmp_path)
+        i.setup(tmp_path, m, script_type="sh")
+
+        skills_dir = tmp_path / ".claude" / "skills"
+        for skill_dir in skills_dir.iterdir():
+            if not skill_dir.is_dir():
+                continue
+            stem = skill_dir.name
+            if stem in ("speckit-clarify", "speckit-checklist"):
+                continue
+            skill_file = skill_dir / "SKILL.md"
+            if skill_file.exists():
+                content = skill_file.read_text(encoding="utf-8")
+                assert "AskUserQuestion" not in content, (
+                    f"{stem}/SKILL.md should not contain AskUserQuestion"
+                )
+
+    def test_clarify_preserves_downstream_behavior(self, tmp_path):
+        """The clarify skill must still contain question-cap and validation instructions."""
+        i = get_integration("claude")
+        m = IntegrationManifest("claude", tmp_path)
+        i.setup(tmp_path, m, script_type="sh")
+
+        skill_file = tmp_path / ".claude" / "skills" / "speckit-clarify" / "SKILL.md"
+        content = skill_file.read_text(encoding="utf-8")
+
+        # 5-question cap
+        assert "5" in content
+        assert "maximum" in content.lower() or "Maximum" in content
+        # Validation pass
+        assert "Validation" in content
+        # Incremental spec writes
+        assert "Clarifications" in content
+
+    def test_checklist_preserves_downstream_behavior(self, tmp_path):
+        """The checklist skill must still contain question-cap and labeling instructions."""
+        i = get_integration("claude")
+        m = IntegrationManifest("claude", tmp_path)
+        i.setup(tmp_path, m, script_type="sh")
+
+        skill_file = tmp_path / ".claude" / "skills" / "speckit-checklist" / "SKILL.md"
+        content = skill_file.read_text(encoding="utf-8")
+
+        # 3-initial / up-to-5-total cap
+        assert "THREE" in content or "three" in content.lower()
+        assert "Q1" in content or "Q4" in content or "Q5" in content
+        # Generation algorithm preserved
+        assert "Generation algorithm" in content
+
+    def test_replace_question_render_block_no_fence(self):
+        """Content without a fence must be returned unchanged."""
+        from specify_cli.integrations.claude import ClaudeIntegration
+
+        content = "No fence here.\nJust regular text.\n"
+        result = ClaudeIntegration.replace_question_render_block(content, "REPLACED")
+        assert result == content
+
+    def test_replace_question_render_block_basic(self):
+        """A fenced block must be replaced with the given content."""
+        from specify_cli.integrations.claude import ClaudeIntegration
+
+        content = (
+            "Before\n"
+            "<!-- speckit:question-render:begin -->\n"
+            "Old content\n"
+            "More old content\n"
+            "<!-- speckit:question-render:end -->\n"
+            "After\n"
+        )
+        result = ClaudeIntegration.replace_question_render_block(content, "NEW BLOCK")
+        assert "NEW BLOCK" in result
+        assert "Old content" not in result
+        assert "Before\n" in result
+        assert "After\n" in result
+        assert "speckit:question-render" not in result
+
+
+class TestNonClaudeIntegrationsParity:
+    """Verify non-Claude integrations produce output unaffected by the fence markers."""
+
+    def test_markdown_integrations_contain_fence_as_is(self, tmp_path):
+        """MarkdownIntegration-based agents must output the fence as plain Markdown."""
+        from specify_cli.integrations import get_integration as get_int
+
+        # Pick a representative Markdown integration
+        for key in ("windsurf", "cursor-agent"):
+            integration = get_int(key)
+            if integration is None:
+                continue
+            m = IntegrationManifest(key, tmp_path)
+            created = integration.setup(tmp_path, m, script_type="sh")
+
+            # Find clarify command output
+            clarify_files = [
+                f for f in created
+                if "clarify" in f.name.lower() or "clarify" in str(f.parent).lower()
+            ]
+            for f in clarify_files:
+                content = f.read_text(encoding="utf-8")
+                # The fence is invisible HTML comments — just verify no
+                # AskUserQuestion was injected
+                assert "AskUserQuestion" not in content, (
+                    f"{key}: {f} should not contain AskUserQuestion"
+                )