ras-commander: Skill for Claude Code

Executing Remote Plans

Use compute_parallel_remote() to distribute HEC-RAS plans across multiple remote machines. Read the primary sources below for complete configuration requirements.

PRIMARY SOURCES (read these for complete details):

• ras_commander/remote/AGENTS.md - Coding conventions, architecture
• .claude/rules/hec-ras/remote.md - Machine setup (Group Policy, Registry, session_id)
• examples/500_remote_execution_psexec.ipynb - Complete PsExec workflow

Quick Start

from ras_commander import init_ras_project, init_ras_worker, compute_parallel_remote

# Initialize project
init_ras_project("/path/to/project", "7.0")

# Create PsExec worker (Windows remote)
worker = init_ras_worker(
    "psexec",
    hostname="192.168.1.100",
    share_path=r"\\192.168.1.100\RasRemote",
    session_id=2,  # CRITICAL: Query with "query session /server:hostname"
    cores_total=16,
    cores_per_plan=4
)

# Execute plans remotely
results = compute_parallel_remote(
    plan_numbers=["01", "02", "03"],
    workers=[worker],
    num_cores=4
)

# Check results
for plan_num, result in results.items():
    if result.success:
        print(f"Plan {plan_num}: SUCCESS ({result.execution_time:.1f}s)")
        print(f"  HDF: {result.hdf_path}")
    else:
        print(f"Plan {plan_num}: FAILED - {result.error_message}")

CRITICAL: Session-Based Execution

HEC-RAS is a GUI application -- always use session-based execution:

worker = init_ras_worker(
    "psexec",
    hostname="192.168.1.100",
    share_path=r"\\192.168.1.100\RasRemote",
    session_id=2,  # CRITICAL: NOT system account
    ...
)

NEVER use system_account=True - HEC-RAS will hang without desktop session.

Determining Session ID

Query from controlling machine:

query session /server:192.168.1.100

# Output:
# SESSIONNAME       USERNAME        ID  STATE
# console           Administrator    2  Active
#                                    ^
#                            Use this value