hello-qm/catgo-lrg/orca

ORCA Router Skill

When to Use

Use this skill when the user requests any ORCA-based calculation. Route to the appropriate sub-skill based on what the user needs:

ORCA-Specific Policies

Method and Basis Set Defaults

Always set software: "orca" in workflow node params. Default method/basis combinations by use case:

Charge and Multiplicity

ORCA requires explicit charge and multiplicity for every calculation. Always ask the user or infer from the structure:

• Neutral closed-shell: charge: 0, multiplicity: 1
• Radical (one unpaired e-): charge: 0, multiplicity: 2
• Anion: charge: -1, multiplicity: 1 (or 2 if radical anion)
• Cation: charge: 1, multiplicity: 1 (or 2 if radical cation)
• Transition metal complexes: determine from d-electron count and ligand field

If unsure, ask the user. Never guess multiplicity for open-shell systems.

Solvent Models

ORCA supports implicit solvation via CPCM:

orca_extra_keywords: "CPCM(Water)"

Common solvents: Water, Acetonitrile, DMSO, THF, Toluene, DCM.

MCP Tool Examples

Check what the user has loaded

catgo_view(action: "get_state")

Always check the viewer state first to confirm a molecular structure is loaded (not a periodic crystal -- ORCA is for molecular calculations).

Create an ORCA optimization workflow

catgo_workflow_engine(action: "create", params: {
  name: "Ethanol B3LYP optimization"
})

Then add an ORCA geo_opt task (see orca/opt skill for full details):

catgo_workflow_engine(action: "add_task", params: {
  workflow_id: "<wf_id>",
  task_type: "geo_opt",
  params: {
    software: "orca",
    orca_method: "B3LYP",
    orca_basis: "def2-SVP",
    charge: 0,
    multiplicity: 1
  }
})

Create an ORCA frequency workflow

catgo_workflow_engine(action: "add_task", params: {
  workflow_id: "<wf_id>",
  task_type: "freq",
  params: {
    software: "orca",
    orca_method: "B3LYP",
    orca_basis: "def2-SVP",
    charge: 0,
    multiplicity: 1
  }
})

Opt + Freq chain (common pattern)

After creating the workflow, add both tasks and connect them:

catgo_workflow_engine(action: "add_task", params: {
  workflow_id: "<wf_id>",
  task_type: "geo_opt",
  task_id: "opt1",
  params: { software: "orca", orca_method: "B3LYP", orca_basis: "def2-SVP",
            charge: 0, multiplicity: 1 }
})

catgo_workflow_engine(action: "add_task", params: {
  workflow_id: "<wf_id>",
  task_type: "freq",
  depends_on: ["opt1"],
  params: { software: "orca", orca_method: "B3LYP", orca_basis: "def2-SVP",
            charge: 0, multiplicity: 1 }
})

Routing Decision Tree

1. Does the user want to find a transition state or reaction barrier?

• YES -> route to orca/neb_ts

1. Does the user want vibrational frequencies, IR spectrum, or thermochemistry?

• YES -> route to orca/freq

1. Does the user want to optimize a geometry or get an energy?

• YES -> route to orca/opt

1. Unsure? Ask the user what property they need.

Common Mistakes

• Using ORCA for periodic systems (ORCA is molecular only -- use VASP or CP2K)
• Forgetting charge/multiplicity (ORCA will fail or give wrong results)
• Using too large a basis for initial screening (start with def2-SVP)
• Not including dispersion for systems with non-covalent interactions