Get shadowmap in HDRP with commandBuffer

The code I've been working on for ages has this line of code

cmd.DrawRenderer(rc.renderer, srcMaterial, rc.submeshIndex, shadowPass);

which it takes the material's shadowpass and put the result into a RenderTexture.

However, the ShadowCaster pass in URP is simple that only have one half type output

 half4 ShadowPassFragment(Varyings input) : SV_TARGET
 {
     Alpha(SampleAlbedoAlpha(input.uv, TEXTURE2D_ARGS(_BaseMap, sampler_BaseMap)).a, _BaseColor, _Cutoff);
     return 0;
 }

And here's the HDRP Lit code for shadow:

 void Frag(  PackedVaryingsToPS packedInput
             #if defined(SCENESELECTIONPASS) || defined(SCENEPICKINGPASS)
             , out float4 outColor : SV_Target0
             #else
                 #ifdef WRITE_MSAA_DEPTH
                 // We need the depth color as SV_Target0 for alpha to coverage
                 , out float4 depthColor : SV_Target0
                     #ifdef WRITE_NORMAL_BUFFER
                     , out float4 outNormalBuffer : SV_Target1
                     #endif
                 #else
                     #ifdef WRITE_NORMAL_BUFFER
                     , out float4 outNormalBuffer : SV_Target0
                     #endif
                 #endif
 
                 // Decal buffer must be last as it is bind but we can optionally write into it (based on _DISABLE_DECALS)
                 #if defined(WRITE_DECAL_BUFFER) && !defined(_DISABLE_DECALS)
                 , out float4 outDecalBuffer : SV_TARGET_DECAL
                 #endif
             #endif
 
             #if defined(_DEPTHOFFSET_ON) && !defined(SCENEPICKINGPASS)
             , out float outputDepth : DEPTH_OFFSET_SEMANTIC
             #endif
         )
 {
     UNITY_SETUP_STEREO_EYE_INDEX_POST_VERTEX(packedInput);
     FragInputs input = UnpackVaryingsToFragInputs(packedInput);
 
     // input.positionSS is SV_Position
     PositionInputs posInput = GetPositionInput(input.positionSS.xy, _ScreenSize.zw, input.positionSS.z, input.positionSS.w, input.positionRWS);
 
 #ifdef VARYINGS_NEED_POSITION_WS
     float3 V = GetWorldSpaceNormalizeViewDir(input.positionRWS);
 #else
     // Unused
     float3 V = float3(1.0, 1.0, 1.0); // Avoid the division by 0
 #endif
 
     SurfaceData surfaceData;
     BuiltinData builtinData;
     GetSurfaceAndBuiltinData(input, V, posInput, surfaceData, builtinData);
 
 #if defined(_DEPTHOFFSET_ON) && !defined(SCENEPICKINGPASS)
     outputDepth = posInput.deviceDepth;
 
 
 #if SHADERPASS == SHADERPASS_SHADOWS
     // If we are using the depth offset and manually outputting depth, the slope-scale depth bias is not properly applied
     // we need to manually apply.
     float bias = max(abs(ddx(posInput.deviceDepth)), abs(ddy(posInput.deviceDepth))) * _SlopeScaleDepthBias;
     outputDepth += bias;
 #endif
 
 #endif
 
 #ifdef SCENESELECTIONPASS
     // We use depth prepass for scene selection in the editor, this code allow to output the outline correctly
     outColor = float4(_ObjectId, _PassValue, 1.0, 1.0);
 #elif defined(SCENEPICKINGPASS)
     outColor = _SelectionID;
 #else
 
     // Depth and Alpha to coverage
     #ifdef WRITE_MSAA_DEPTH
         // In case we are rendering in MSAA, reading the an MSAA depth buffer is way too expensive. To avoid that, we export the depth to a color buffer
         depthColor = packedInput.vmesh.positionCS.z;
 
         #ifdef _ALPHATOMASK_ON
         // Alpha channel is used for alpha to coverage
         depthColor.a = SharpenAlpha(builtinData.opacity, builtinData.alphaClipTreshold);
         #endif
     #endif
 
     #if defined(WRITE_NORMAL_BUFFER)
     EncodeIntoNormalBuffer(ConvertSurfaceDataToNormalData(surfaceData), outNormalBuffer);
     #endif
 
     #if defined(WRITE_DECAL_BUFFER) && !defined(_DISABLE_DECALS)
     DecalPrepassData decalPrepassData;
     // We don't have the right to access SurfaceData in a shaderpass.
     // However it would be painful to have to add a function like ConvertSurfaceDataToDecalPrepassData() to every Material to return geomNormalWS anyway
     // Here we will put the constrain that any Material requiring to support Decal, will need to have geomNormalWS as member of surfaceData (and we already require normalWS anyway)
     decalPrepassData.geomNormalWS = surfaceData.geomNormalWS;
     decalPrepassData.decalLayerMask = GetMeshRenderingDecalLayer();
     EncodeIntoDecalPrepassBuffer(decalPrepassData, outDecalBuffer);
     #endif
 
 #endif // SCENESELECTIONPASS
 }

Can anybody please tell me which one in HDRP is equivalent to the URP version, and how am I going to get the same shadowmap for the command buffer as URP?