Fix rendering issue in documentation. (#2687)

* Fix rendering issue in documentation

* Add prompts for NoahmpWriteLand

Author: akashdhruv

Docs/sphinx_doc/CouplingToNoahMP.rst

@@ -118,15 +118,16 @@ to configure your LLM environment. You will need API access for your preferred
 model (e.g., OpenAI, Argo, etc.). Tutorials are available at
 `https://github.com/akashdhruv/codescribe-tutorial <https://github.com/akashdhruv/codescribe-tutorial>`_.
 
-1. Edit the prompt file ``prompts/noahmpio_update.toml`` to specify which
+1. Edit the prompt file **prompts/noahmpio_update.toml** to specify which
    variables should be exposed to the C++ interface.
-2. Run the following commands to generate or update bindings::
 
-   # In Submodules/Noah-MP/drivers/erf
+2. Run the following commands to generate or update bindings in **Submodules/Noah-MP/drivers/erf**::
+
    code-scribe update NoahmpIO.H NoahmpIO.cpp NoahmpIO_fi.F90 \
        -p prompts/noahmpio_update.toml -m <openai|argo-gpt4o|...>
 
-   # In Source/LandSurfaceModel/Noah-MP
+3. Run the following to generate or update bindings in **Source/LandSurfaceModel/Noah-MP**::
+
    code-scribe update ERF_NOAHMP.cpp \
        -p prompts/noahmpio_update.toml -m <openai|argo-gpt4o|...>
 
@@ -138,9 +139,7 @@ with::
 
    real(kind=C_DOUBLE)
 
-This ensures compatibility with the C++ side.
-
-Alternatively, CodeScribe can perform this update automatically (depending on your
+This ensures compatibility with the C++ side. Alternatively, CodeScribe can perform this update automatically (depending on your
 model’s context length) using::
 
    code-scribe update NoahmpIOVarType.F90 \

Source/LandSurfaceModel/Noah-MP/prompts/noahmpio_update.toml

@@ -1,5 +1,8 @@
+# Usage: code-scribe update ERF_NOAHMP.cpp \
+#                           -p prompts/noahmpio_update.toml -m <openai|argo|..>
+
 [[chat.user]]
-content = """
+content = '''
 You are a code-update assistant for extending the ERF-NoahMP C++ driver interface.
 
 Only edit the appended files. Focus on the coupling logic inside
@@ -12,23 +15,23 @@ patterns used for existing variables such as `U_PHY`, `V_PHY`, `T_PHY`, `QV_CURR
 
 1. Context
 
-- The ERF-NoahMP coupling exchanges field data via `noahmpio_vect`, a per-tile
-  vector of `NoahmpIO_type` objects. Each object corresponds to a domain tile
-  in `cons_in`.
+-  The ERF-NoahMP coupling exchanges field data via `noahmpio_vect`, a per-tile
+   vector of `NoahmpIO_type` objects. Each object corresponds to a domain tile
+   in `cons_in`.
 
-- Each time step, `Advance_With_State()` performs the following sequence:
-    (1) Extract 3D Array4<Real> views from ERF state variables.
-    (2) Allocate temporary host-accessible FArrayBox buffers (The_Pinned_Arena()).
-    (3) Launch ParallelFor GPU kernels to copy ERF data into pinned buffers.
-    (4) Synchronize GPU streams, then perform CPU loops (LoopOnCpu) to copy
-        those buffers into corresponding NoahMP fields (noahmpio->VAR(i,k,j)).
-    (5) Call the NoahMP driver (DriverMain()).
-    (6) Copy computed fluxes and surface variables back from NoahMP to ERF arrays.
+-  Each time step, `Advance_With_State()` performs the following sequence:
+   (1) Extract 3D Array4<Real> views from ERF state variables.
+   (2) Allocate temporary host-accessible FArrayBox buffers (The_Pinned_Arena()).
+   (3) Launch ParallelFor GPU kernels to copy ERF data into pinned buffers.
+   (4) Synchronize GPU streams, then perform CPU loops (LoopOnCpu) to copy
+         those buffers into corresponding NoahMP fields (noahmpio->VAR(i,k,j)).
+   (5) Call the NoahMP driver (DriverMain()).
+   (6) Copy computed fluxes and surface variables back from NoahMP to ERF arrays.
 
-- Newly added variables follow the same pattern:
-  - GPU->Host copy -> LoopOnCpu -> assign to noahmpio->NEWVAR
-  - DriverMain() executes the Fortran driver
-  - Host->GPU copy -> ParallelFor -> assign to ERF Array4<Real> fields
+-  Newly added variables follow the same pattern:
+   - GPU->Host copy -> LoopOnCpu -> assign to noahmpio->NEWVAR
+   - DriverMain() executes the Fortran driver
+   - Host->GPU copy -> ParallelFor -> assign to ERF Array4<Real> fields
 
 2. Task Details
 
@@ -37,53 +40,53 @@ variables RAINLSM and any other surface fields needed for the ERF side.
 
 Changes Required
 
-- Host buffers
+-  Host buffers
 Add new temporary FArrayBox allocations for each new variable:
-    FArrayBox tmp_rainlsm(bx, 1, The_Pinned_Arena());
+   FArrayBox tmp_rainlsm(bx, 1, The_Pinned_Arena());
 
 and retrieve their array handles:
-    auto const& tmp_rainlsm_arr = tmp_rainlsm.array();
+   auto const& tmp_rainlsm_arr = tmp_rainlsm.array();
 
-- Device->Host copy (ParallelFor)
+-  Device->Host copy (ParallelFor)
 If the new field originates on the ERF side (e.g., forcing input), copy it from
 ERF arrays into the pinned buffer in the existing ParallelFor block.
 
 If it is computed by NoahMP, skip this step and handle it in the post-DriverMain()
 copy-back section.
 
-- Host->NoahMP copy (LoopOnCpu)
+-  Host->NoahMP copy (LoopOnCpu)
 After GPU synchronization (Gpu::streamSynchronize()), extend the LoopOnCpu
 block that writes to noahmpio fields:
-    noahmpio->RAINLSM(i,j) = tmp_rainlsm_arr(i,j,0);
+   noahmpio->RAINLSM(i,j) = tmp_rainlsm_arr(i,j,0);
 
-- Post-DriverMain copy (LoopOnCpu)
+-  Post-DriverMain copy (LoopOnCpu)
 Add the reverse direction: copying noahmpio->RAINLSM values back into the
 temporary FArrayBox before GPU upload:
-    h_rainlsm_arr(i,j,0) = noahmpio->RAINLSM(i,j);
+   h_rainlsm_arr(i,j,0) = noahmpio->RAINLSM(i,j);
 
-- Host->Device copy-back (ParallelFor)
+-  Host->Device copy-back (ParallelFor)
 Finally, assign results into the correct ERF field:
-    RAINLSM(i,j,0) = tmp_rainlsm_arr(i,j,0);
+   RAINLSM(i,j,0) = tmp_rainlsm_arr(i,j,0);
 
 Follow the same block and comment style used for variables like TSK, EMISS,
 and HFX.
 
 3. Validation
 
-- Verify that array ranks match (2D for surface variables, 3D for atmospheric variables).
-- Ensure the order of CPU/GPU synchronization and host copies is preserved.
-- Confirm that noahmpio_vect indexing (idb) and bounds (bx.smallEnd, domain.smallEnd)
-  remain untouched.
-- Maintain one-to-one consistency between ERF arrays and NoahMP I/O variables.
+-  Verify that array ranks match (2D for surface variables, 3D for atmospheric variables).
+-  Ensure the order of CPU/GPU synchronization and host copies is preserved.
+-  Confirm that noahmpio_vect indexing (idb) and bounds (bx.smallEnd, domain.smallEnd)
+   remain untouched.
+-  Maintain one-to-one consistency between ERF arrays and NoahMP I/O variables.
 
 4. Actual Task
 
 Modify and link the following variable declared in NoahmpIOVarType.F90 and
 allocated in NoahmpIOVarInitMod.F90:
 
-- RAINLSM, 2D array of type real(kind=kind_noahmp), with bounds (XSTART:XEND, YSTART:YEND)
+-  RAINLSM, 2D array of type real(kind=kind_noahmp), with bounds (XSTART:XEND, YSTART:YEND)
 
 Ensure the field is fully integrated into the ERF-NoahMP coupling sequence
 inside Advance_With_State() following the memory flow used for TSK.
 
-"""
+'''