Document config-driven pipeline with customisable parameters in README

Replace the old pipeline section with comprehensive documentation of
the YAML config system, parameter tables, Docker-based SU2 execution,
and system-level dependency notes. Add .gitignore.

Made-with: Cursor

Author: Kugel-Blitz-13

.gitignore

@@ -205,3 +205,4 @@ cython_debug/
 marimo/_static/
 marimo/_lsp/
 __marimo__/
+.DS_Store

README.md

@@ -90,23 +90,68 @@ To install SU2 alongside the MCP server, opt in to the extra dependency group wh
 pip install su2-mcp-server[su2]
 ```
 
-### TiGL → SU2 pipeline (external agent)
+### TiGL → SU2 → pyCycle pipeline
 
-An external agent can chain TiGL MCP, SU2 MCP, and (optionally) Pycycle MCP to design and analyze an aircraft:
+An orchestration script (`pipeline/tigl_to_su2.py`) chains all three MCP servers into a single automated run:
 
-1. **TiGL MCP**: Open CPACS, export configuration as STEP (base64).
-2. **SU2 MCP**: `create_su2_session` → `generate_mesh_from_step(session_id, step_base64)` (requires `gmsh` on PATH) → `update_config_entries` to set Euler/freestream and `MARKER_FAR`/`MARKER_EULER` to match the mesh markers (FARFIELD, WALL) → `run_su2_solver` → `list_result_files` / `read_history_csv` / `get_result_file_base64` for results.
-3. **Pycycle MCP**: Use for propulsion/cycle analysis as needed.
+```
+CPACS XML → [TiGL MCP] → STEP → [Gmsh] → SU2 mesh → [SU2] → CL/CD → [pyCycle MCP] → TSFC, fuel flow
+```
 
-For a single Docker environment that has TiGL, Gmsh, and SU2, use the `tigl-mcp` image (which includes SU2 and Gmsh) and run the SU2 MCP server in that same environment so `generate_mesh_from_step` can call `gmsh`.
+1. **TiGL MCP** (port 8000): Open CPACS, export configuration as watertight STEP.
+2. **SU2 MCP / Gmsh**: Volume mesh from STEP (BooleanFragments, FARFIELD/WALL markers), then Euler CFD to get CL/CD.
+3. **pyCycle MCP** (port 8001): Flight conditions + drag → engine sizing (thrust, TSFC, fuel flow).
 
-### Running the server when SU2 is not installed locally
+#### Config-driven runs
 
-The server runs `SU2_CFD` when you call `run_su2_solver`. If SU2 is not on your machine, run the SU2 MCP server inside a container that has SU2 (e.g. the [tigl-mcp](https://github.com/cmudrc/tigl-mcp) Docker image). From the host, mount this repo and start the server:
+The pipeline uses a YAML config file (`pipeline_config.yaml`) that controls all ~15 user-relevant parameters. CLI arguments override config values.
 
 ```bash
-docker run --rm -p 8002:8002 -v /path/to/su2-mcp:/app/su2-mcp tigl-mcp:dev \
-  bash -lc "pip install -e /app/su2-mcp && su2-mcp-server --transport http --host 0.0.0.0 --port 8002 --path /mcp"
+# Run with config file
+python pipeline/tigl_to_su2.py D150.xml --config pipeline/pipeline_config.yaml -o output/
+
+# Override specific values via CLI
+python pipeline/tigl_to_su2.py D150.xml -c pipeline/pipeline_config.yaml --mach 0.85 --density 80
+
+# Defaults only (no config file needed)
+python pipeline/tigl_to_su2.py D150.xml -o output/
 ```
 
-Then clients can connect to `http://localhost:8002/mcp`. For a pre-built script that does this, see the parent repo that contains both tigl-mcp and su2-mcp.
+Config file sections: `flight` (mach, aoa, altitude), `aircraft` (ref_area, ref_length), `meshing` (surface_density, farfield_factor, algorithm), `cfd` (iterations, CFL, limiter), `engine` (type, default thrust). A `resolved_config.yaml` is saved alongside outputs for reproducibility.
+
+#### Customisable parameters
+
+| Category | Key parameters | Default |
+|----------|---------------|---------|
+| Flight | mach, aoa, altitude | 0.78, 2°, 35,000 ft |
+| Mesh quality | `surface_density` (span/N) | 30 (coarse) — set 80-150 for smooth |
+| CFD | iterations, CFL, limiter_coeff | 250, 1.0, 0.1 |
+| Engine | type, default_thrust_lbf | turbofan, 5900 |
+
+#### Hardcoded (not exposed via config)
+
+- Solver: Euler (inviscid) — RANS support is a future milestone
+- Flux scheme: ROE with Venkatakrishnan limiter
+- pyCycle engine cycle internals (duct losses, bleed fractions, nozzle Cv)
+- TiGL STEP deflection tolerance (0.001)
+
+### Running SU2 when not installed locally
+
+SU2 is available inside the [tigl-mcp](https://github.com/cmudrc/tigl-mcp) Docker image. Copy your mesh and config into the running container:
+
+```bash
+docker cp output/aircraft_volume.su2 <container>:/tmp/aircraft_volume.su2
+docker cp output/euler.cfg <container>:/tmp/euler.cfg
+docker exec <container> bash -c "cd /tmp && conda run -n tigl SU2_CFD euler.cfg"
+docker cp <container>:/tmp/history.csv output/
+docker cp <container>:/tmp/vol_solution.vtu output/
+```
+
+### System-level dependencies
+
+These cannot be pip-installed and require conda or system packages:
+
+- **SU2** (`SU2_CFD`): `conda install -c conda-forge su2` or use the Docker image
+- **Gmsh** (Python API): `pip install gmsh` (or `conda install -c conda-forge gmsh`)
+
+The MCP server itself (`pip install .`) is pure Python and installs cleanly.