aiida_mcp.plugins.workflow module¶
Generic describe and submit helpers shared by workflow (WorkChain) plugins.
Unlike calcjobs, WorkChains commonly build their input namespace via spec.expose_inputs(…) from other processes, producing deeply nested PortNamespace inputs rather than a flat set of ports. Generically flattening and describing that tree would be misleading (most of a WorkChain’s real inputs would appear as opaque, skipped namespaces), so describe_workflow only reports outputs and exit codes, not inputs. A WorkChain’s own submit_* tool and its hand-written docstring are the authoritative description of what it actually accepts to submit.
Some WorkChains (e.g. one that simply nests a calcjob’s own inputs under one extra namespace) still take inputs via a plain, JSON-shaped dict, just like a calcjob’s submit_calcjob; submit_workflow supports that case. Others build their own ProcessBuilder via a bespoke get_builder_from_protocol-style classmethod (e.g. one that orchestrates several other, heavier WorkChains); there is no generic helper for that case — wrap the classmethod directly.
- aiida_mcp.plugins.workflow.describe_workflow(entry_point: str) dict[str, Any]¶
Describe the plugin-specific outputs and exit codes of a WorkChain.
Outputs and exit codes inherited unchanged from the base WorkChain are left out.
- Parameters:
entry_point (str) – The workflow entry point name, e.g.
cif2j.wannier_tb2j.- Returns:
The workflow’s entry point, class, description, plugin-specific outputs and exit codes.
- Return type:
dict[str, Any]
- Raises:
ToolError – If entry_point is not installed.
- aiida_mcp.plugins.workflow.submit_workflow(entry_point: str, inputs: dict[str, Any]) ProcessNode¶
Validate inputs against the workflow spec, store the data nodes and submit.
Nothing is stored if validation fails. Inputs for an exposed namespace (e.g. from spec.expose_inputs(OtherProcess, namespace=”other”)) are given as a nested dict under that namespace’s key, exactly as they would be passed to submit/run directly; any orm.Data values nested inside such a dict are stored too.
- Parameters:
entry_point (str) – The workflow entry point name, e.g.
chemshell.opt.inputs (dict[str, Any]) – The workflow’s inputs, including
metadata(nested under any namespace that exposes it). Unstored orm.Data values, including ones nested inside namespace dicts, are stored before submission.
- Returns:
The submitted process node.
- Return type:
orm.ProcessNode
- Raises:
ToolError – If entry_point is not installed, or inputs fails validation.
- aiida_mcp.plugins.workflow.workflow_available(entry_point: str) bool¶
Return whether a workflow entry point is installed.
- Parameters:
entry_point (str) – The workflow entry point name, e.g.
cif2j.wannier_tb2j.- Returns:
True if the entry point is installed.
- Return type:
bool