Hello-QM/catgo-LRG / server/catgo/workflow
catgo-workflow
VerifiedCreate and manage computational chemistry workflows with CatGo. Supports VASP, CP2K, ORCA, MLP, LAMMPS. Build OER/HER/CO2RR workflows, geometry optimization, frequency analysis, Gibbs energy calculations.
mcp_tools.py
View on GitHub"""MCP tool interface for AI agents to create/manage CatGo workflows.
AI agents call these via MCP protocol. Each action maps to service.py.
"""
from __future__ import annotations
import asyncio
from typing import Any
import contextvars
from catgo.workflow.task_ids import make_task_id
# Track the active project context (e.g., when called from ProjectDashboard)
_active_project_id: contextvars.ContextVar[str | None] = contextvars.ContextVar('active_project_id', default=None)
def set_active_project(project_id: str | None) -> None:
"""Set the active project context for workflow creation."""
_active_project_id.set(project_id)
def get_active_project() -> str | None:
"""Get the active project context."""
return _active_project_id.get()
def get_tool_definition() -> dict:
"""Return the MCP tool schema for catgo_workflow_engine."""
return {
"name": "catgo_workflow_engine",
"description": (
"Create and manage computational chemistry workflows. Build: create -> "
"add_task (one per node) -> submit. "
"Actions: create, add_task, submit, status, list, modify_params, retry, "
"pause, resume, reset, get_result, get_dag. "
"WIRE TASKS by data flow (there is no separate connect action): pass an "
"upstream task's output as an input param of the downstream add_task, using "
"an output reference {'_ref': '<upstream_task_id>', '_key': '<output_key>'}. "
"e.g. add_task geo_opt with params {'structure': {'_ref': '<slab_task_id>', "
"'_key': 'structure'}} creates the slab.structure -> geo_opt.structure edge. "
"add_task returns the new task_id. "
"IMPORTANT: Before using action='submit', you MUST ask the user which HPC "
"cluster to use (e.g., Expanse, Shaheen, local) and confirm job parameters "
"(partition, account, walltime, ntasks). Never submit without user confirmation. "
"ALSO confirm the pseudopotential/POTCAR directory for that cluster: if you are "
"not certain where POTCAR/pseudopotential files live (potcar_root etc.), STOP and "
"ASK THE USER -- do NOT guess (a wrong path fails every job; it is per-user/"
"per-cluster and not inferable from another workflow). On Expanse POTCAR can be "
"generated via 'echo -e 103 | vaspkit'. SAME for the compute binary: confirm how "
"the executable is loaded/invoked (vasp_command + 'module load'/'conda activate'/"
"full path); if not certain, ASK THE USER -- a wrong command/module dies with "
"'command not found' (e.g. execve(): vasp_std: No such file or directory). "
"HPC job parameters can be set per-task via add_task params. "
"Task ids are namespaced as '{workflow_id}:{node_id}'. For per-task actions "
"(get_result, modify_params, retry) pass either an explicit 'task_id', or "
"'workflow_id' + 'node_id' (the graph node id) which will be namespaced for you."
),
"inputSchema": {
"type": "object",
"properties": {
"action": {
"type": "string",
"enum": [
"create", "add_task", "submit",
"status", "list", "modify_params", "retry",
"pause", "resume", "reset", "get_result", "get_dag",
],
},
"params": {
"type": "object",
"description": (
"Action-specific parameters. add_task: workflow_id, task_type, and "
"input params (use {'_ref':'<task_id>','_key':'<key>'} values to wire "
"from upstream tasks). For per-task actions (get_result, "
"modify_params, retry), provide either 'task_id' (already namespaced "
"as '{workflow_id}:{node_id}') or 'workflow_id' + 'node_id'."
),
},
},
"required": ["action"],
},
}
def _resolve_task_id(params: dict[str, Any]) -> str:
"""Resolve a (namespaced) task id from MCP params.
Task ids are namespaced ``{workflow_id}:{node_id}`` (issue #227). Result/status
tools may be called either with an explicit ``task_id`` (back-compat) or with
``workflow_id`` + ``node_id`` (the graph node id), which are namespaced here.
"""
if params.get("task_id"):
return params["task_id"]
wf = params.get("workflow_id")
node = params.get("node_id")
if wf and node:
return make_task_id(wf, node)
raise ValueError("provide task_id, or workflow_id + node_id")
def _get_db():
"""Get a WorkflowDB instance from config."""
from catgo.workflow.db import WorkflowDB
from catgo.workflow.config import load_config
from pathlib import Path
config = load_config()
db_path = str(Path(config["paths"]["db_path"]).expanduser())
return WorkflowDB(db_path)
def _dispatch_sync(action: str, params: dict[str, Any]) -> dict[str, Any]:
"""Synchronous dispatch. Runs in a worker thread via asyncio.to_thread."""
from catgo.workflow import service
db = _get_db()
if action == "create":
wf = service.create_workflow(db, params.get("name", "Untitled"), params.get("config"))
# Assign to project if provided explicitly OR from active context
project_id = params.get("project_id") or get_active_project()
if project_id:
db.assign_project(wf["workflow_id"], project_id)
wf["project_id"] = project_id
return wf
elif action == "add_task":
nested = params.get("params")
if isinstance(nested, dict):
task_params = dict(nested)
else:
task_params = {k: v for k, v in params.items()
if k not in ("workflow_id", "task_type", "name", "system_name", "params")}
return service.add_task(
db, params["workflow_id"], params["task_type"],
name=params.get("name"), system_name=params.get("system_name"),
**task_params,
)
elif action == "submit":
return service.submit(db, params["workflow_id"])
elif action == "status":
return service.get_status(db, params["workflow_id"])
elif action == "list":
return {"workflows": service.list_workflows(db)}
elif action == "get_dag":
return db.get_dag(params["workflow_id"])
elif action == "get_result":
result = db.get_result(_resolve_task_id(params))
return result or {"error": "No result found"}
elif action == "modify_params":
return service.modify_task_params(db, _resolve_task_id(params), params.get("updates", {}))
elif action == "retry":
reset_ids = service.retry_task(db, _resolve_task_id(params))
return {"reset_tasks": reset_ids}
elif action == "pause":
return service.pause(db, params["workflow_id"])
elif action == "resume":
return service.resume(db, params["workflow_id"])
elif action == "reset":
return service.reset(db, params["workflow_id"])
else:
return {"error": f"Unknown action: {action}"}
async def handle_tool_call(action: str, params: dict[str, Any]) -> dict[str, Any]:
"""Route MCP tool calls to service functions.
Offloads sync SQLite / service work to a thread so the FastAPI event loop
never blocks on DB I/O, HPC SSH, or file system calls.
"""
return await asyncio.to_thread(_dispatch_sync, action, params)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |