SU

subscription-sync

Manages synchronization lifecycles and event buffering for real-time frontend applications.

Install

mkdir -p .claude/skills/subscription-sync && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/18225" && unzip -o skill.zip -d .claude/skills/subscription-sync && rm skill.zip

Installs to .claude/skills/subscription-sync

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Real-time data synchronization patterns for ChainGraph frontend. Use when working on WebSocket subscriptions, event buffers, tRPC subscriptions, flow synchronization, or execution event streaming. Covers subscription lifecycle, event buffering, race condition solutions. Triggers: subscription, sync, real-time, websocket, event buffer, tRPC subscription, flow events, onData, patronum interval.
395 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Advanced

Key capabilities

  • Manage real-time data synchronization via WebSocket subscriptions
  • Utilize two separate tRPC clients for flow editing and execution events
  • Implement event buffering to prevent race conditions during flow initialization
  • Handle flow subscription lifecycle including connecting, subscribed, and disconnected states
  • Subscribe to execution events and ensure proper sequencing

How it works

The skill uses two tRPC clients for separate WebSocket connections to handle flow editing and execution events. It employs an event buffer to batch and flush flow events atomically, preventing race conditions.

Inputs & outputs

You give it
Real-time data streams from ChainGraph backend
You get back
Synchronized frontend state in Effector Stores

When to use subscription-sync

  • Debug real-time data sync
  • Implement WebSocket subscription
  • Manage event buffer state

About this skill

Subscription Sync Patterns

This skill covers the real-time data synchronization system between ChainGraph backend and frontend via WebSocket subscriptions.

Architecture Overview

┌──────────────────────────────────────────────────────────────┐
│                      BACKEND (tRPC)                           │
│                                                               │
│  Flow Subscription           Execution Subscription           │
│  ├─ FlowInitStart            ├─ EXECUTION_CREATED            │
│  ├─ NodesAdded               ├─ FLOW_STARTED                 │
│  ├─ EdgesAdded               ├─ NODE_STARTED                 │
│  ├─ FlowInitEnd              ├─ NODE_COMPLETED               │
│  ├─ NodeUpdated              ├─ EDGE_TRANSFER                │
│  ├─ PortUpdated              └─ FLOW_COMPLETED               │
│  └─ ...                                                       │
└──────────────────┬───────────────────────┬───────────────────┘
                   │ WebSocket             │ WebSocket
                   ▼                       ▼
┌──────────────────────────────────────────────────────────────┐
│                      FRONTEND                                  │
│                                                               │
│  ┌─────────────────────┐    ┌─────────────────────┐          │
│  │ $trpcClient         │    │ $trpcClientExecutor │          │
│  │ ws://localhost:3001 │    │ ws://localhost:4021 │          │
│  └──────────┬──────────┘    └──────────┬──────────┘          │
│             │                          │                      │
│             ▼                          ▼                      │
│  ┌─────────────────────┐    ┌─────────────────────┐          │
│  │ Flow Event Buffer   │    │ Execution Events    │          │
│  │ (50ms batching)     │    │ (direct processing) │          │
│  └──────────┬──────────┘    └──────────┬──────────┘          │
│             │                          │                      │
│             ▼                          ▼                      │
│  ┌────────────────────────────────────────────────┐          │
│  │              Effector Stores                    │          │
│  │  $nodes, $edges, $portValues, $execution        │          │
│  └────────────────────────────────────────────────┘          │
└──────────────────────────────────────────────────────────────┘

Two tRPC Clients

ChainGraph frontend maintains TWO separate WebSocket connections:

Files:

  • Main Server Client: apps/chaingraph-frontend/src/store/trpc/store.ts
  • Executor Server Client: apps/chaingraph-frontend/src/store/trpc/execution-client.ts
// Main Server - Flow editing operations (store.ts)
export const $trpcClient = trpcDomain.createStore<TRPCClient | null>(null)
// Connects to: ws://localhost:3001

// Executor Server - Execution events (execution-client.ts)
export const $trpcClientExecutor = trpcDomain.createStore<TRPCClient | null>(null)
// Connects to: ws://localhost:4021

Why Two Clients?

  1. Separation of Concerns: Flow editing and execution are independent
  2. Load Distribution: Heavy execution traffic doesn't block editing
  3. Independent Scaling: Executor can scale separately
  4. Failure Isolation: Execution server crash doesn't break editing

Flow Subscription Lifecycle

Files:

  • Subscription: apps/chaingraph-frontend/src/store/flow/subscription.ts
  • Event Buffer: apps/chaingraph-frontend/src/store/flow/event-buffer.ts

Event Sequence

1. FlowInitStart
   └─ Clear existing nodes/edges
   └─ Set status: CONNECTING → SUBSCRIBED

2. NodesAdded (batch)
   └─ Buffer accumulates events

3. EdgesAdded (batch)
   └─ Buffer accumulates events

4. FlowInitEnd (COMMIT SIGNAL)
   └─ Buffer flushes immediately
   └─ All events processed atomically
   └─ Nodes render BEFORE edges (race condition solved)

5. Live Updates (ongoing)
   └─ Buffer with 50ms interval
   └─ NodeUpdated, PortUpdated, EdgeAdded, etc.

Subscription Status

enum FlowSubscriptionStatus {
  IDLE = 'idle',
  CONNECTING = 'connecting',
  SUBSCRIBED = 'subscribed',
  ERROR = 'error',
  DISCONNECTED = 'disconnected',
}

Event Buffer Pattern

Problem: Race condition where edges render before nodes during flow initialization.

Root Cause:

1. addNodes triggers xyflowStructureChanged with 50ms debounce
2. setEdges updates $xyflowEdges immediately
3. $xyflowEdges filters out edges because $xyflowNodes is empty

Solution: Buffer ALL FlowEvents and flush atomically on FlowInitEnd.

File: apps/chaingraph-frontend/src/store/flow/event-buffer.ts

import { interval } from 'patronum'

// Buffer accumulates events
export const $flowEventBuffer = flowDomain.createStore<FlowEvent[]>([])
  .on(flowEventReceived, (buffer, event) => [...buffer, event])

// Ticker runs every 50ms (configurable via VITE_FLOW_EVENT_BUFFER_INTERVAL)
const ticker = interval({
  timeout: 50,  // BUFFER_INTERVAL_MS
  start: tickerStart,
  stop: tickerStop,
})

// Auto-start ticker when first event arrives
sample({
  clock: flowEventReceived,
  source: $flowEventBuffer,
  filter: buffer => buffer.length === 1,  // Buffer was empty
  target: tickerStart,
})

// Auto-stop ticker when buffer is empty
sample({
  clock: $flowEventBuffer,
  filter: buffer => buffer.length === 0,
  target: tickerStop,
})

// CRITICAL: Flush immediately on FlowInitEnd
sample({
  clock: flowEventReceived,
  filter: event => event.type === FlowEventType.FlowInitEnd,
  target: flushBuffer,
})

Buffer Processing Flow

Subscription → flowEventReceived → $flowEventBuffer
                                         │
                    ┌────────────────────┴────────────────────┐
                    │                                          │
              [FlowInitEnd]                               [50ms tick]
                    │                                          │
                    ▼                                          ▼
             flushBuffer (immediate)              processBufferFx (batched)
                    │                                          │
                    └────────────────┬─────────────────────────┘
                                     │
                                     ▼
                              newFlowEvents (batch of FlowEvent[])
                                     │
                                     ▼
                              Event Handlers in stores.ts

Execution Subscription

File: apps/chaingraph-frontend/src/store/execution/subscription.ts

Execution events are processed directly (no buffering needed):

// Subscribe to execution events
// Note: No .execution namespace - procedures are at router root
const subscription = trpcClientExecutor.subscribeToExecutionEvents.subscribe(
  { executionId, fromIndex: 0 },
  {
    onData: (event) => {
      executionEventReceived(event)  // Direct dispatch
    },
    onError: (error) => {
      executionError(error)
    },
  }
)

Execution Event Types

enum ExecutionEventEnum {
  EXECUTION_CREATED = 'EXECUTION_CREATED',  // index -1
  FLOW_STARTED = 'FLOW_STARTED',
  NODE_STARTED = 'NODE_STARTED',
  NODE_COMPLETED = 'NODE_COMPLETED',
  NODE_FAILED = 'NODE_FAILED',
  EDGE_TRANSFER_COMPLETED = 'EDGE_TRANSFER_COMPLETED',
  FLOW_COMPLETED = 'FLOW_COMPLETED',
  FLOW_FAILED = 'FLOW_FAILED',
  CHILD_EXECUTION_SPAWNED = 'CHILD_EXECUTION_SPAWNED',
}

Key Files

FilePurpose
src/store/trpc/store.tstRPC client stores
src/store/flow/subscription.tsFlow subscription management
src/store/flow/event-buffer.tsEvent buffering with patronum
src/store/execution/subscription.tsExecution event subscription
src/store/flow/stores.tsEvent handlers (newFlowEvents)

Common Patterns

Subscribe to Flow

import { subscribeToFlowFx, unsubscribeFromFlowFx } from '@/store/flow/subscription'

// Subscribe
subscribeToFlowFx(flowId)

// Unsubscribe (cleanup)
unsubscribeFromFlowFx()

Handle Flow Events

// In stores.ts
sample({
  clock: newFlowEvents,
  filter: events => events.some(e => e.type === FlowEventType.NodeUpdated),
  fn: events => events.filter(e => e.type === FlowEventType.NodeUpdated),
  target: processNodeUpdates,
})

Subscribe to Execution

import { subscribeToExecutionFx } from '@/store/execution/subscription'

// Subscribe and wait for EXECUTION_CREATED
await subscribeToExecutionFx({ executionId })

// Start execution after subscription is ready
startExecution({ executionId })

Anti-Patterns

Anti-Pattern #1: Processing events without buffering

// ❌ BAD: Direct dispatch causes race conditions
onData: (event) => {
  newFlowEvents([event])  // Edges may render before nodes!
}

// ✅ GOOD: Use buffer
onData: (event) => {
  flowEventReceived(event)  // Buffer handles ordering
}

Anti-Pattern #2: Not waiting for EXECUTION_CREATED

// ❌ BAD: Start before subscription is ready
startExecution({ executionId })
subscribeToExecutionFx({ executionId })  // Might miss events!

// ✅ GOOD: Subscribe first, then start
await subscribeToExecutionFx({ executionId })
startExecution({ executionId })

Anti-Pattern #3: Not cleaning up subscriptions

// ❌ BAD: Memory leak
useEffect(() => {
  subscribeToFlowFx(flowId)
  // No cleanup!
}, [flowId])

// ✅ GOOD: Cleanup on unmount/change
useEffect(() => {
  subscribeToFlowFx(flowId)
  return () => {
    unsubscribeFromFlowFx()
  }
}, [flowId])

Quick Reference

NeedPatternFile
Subscribe to flowsubscribeToFlowFx(flowId)flow/subscription.ts
Buffer eventsflowEventReceived(event)flow/event-buffer.ts
Process buffered eventsnewFlowEvents eventflow/stores.ts
Subscribe to execution`subs

Content truncated.

When not to use it

  • When processing events without buffering, which causes race conditions
  • When starting execution before the subscription is ready
  • When not cleaning up subscriptions, which causes memory leaks

Limitations

  • Requires two separate WebSocket connections for flow editing and execution
  • Relies on specific event buffering patterns to prevent race conditions
  • Requires explicit cleanup of subscriptions to avoid memory leaks

How it compares

This skill provides specific patterns and architecture for managing real-time data synchronization with separate clients and event buffering, which is more structured than a generic WebSocket implementation.

Compared to similar skills

subscription-sync side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
subscription-sync (this skill)06moNo flagsAdvanced
telegram-mini-app626moReviewAdvanced
create-mcp-app34moReviewAdvanced
shopify-apps13moReviewIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

telegram-mini-app

davila7

Expert in building Telegram Mini Apps (TWA) - web apps that run inside Telegram with native-like experience. Covers the TON ecosystem, Telegram Web App API, payments, user authentication, and building viral mini apps that monetize. Use when: telegram mini app, TWA, telegram web app, TON app, mini app.

62163

create-mcp-app

modelcontextprotocol

This skill should be used when the user asks to "create an MCP App", "add a UI to an MCP tool", "build an interactive MCP View", "scaffold an MCP App", or needs guidance on MCP Apps SDK patterns, UI-resource registration, MCP App lifecycle, or host integration. Provides comprehensive guidance for building MCP Apps with interactive UIs.

331

shopify-apps

alinaqi

Shopify app development - Remix, Admin API, checkout extensions

19

ccxt-typescript

ccxt

CCXT cryptocurrency exchange library for TypeScript and JavaScript developers (Node.js and browser). Covers both REST API (standard) and WebSocket API (real-time). Helps install CCXT, connect to exchanges, fetch market data, place orders, stream live tickers/orderbooks, handle authentication, and manage errors. Use when working with crypto exchanges in TypeScript/JavaScript projects, trading bots, arbitrage systems, or portfolio management tools. Includes both REST and WebSocket examples.

15

add-nodebridge-handler

neovateai

Use this skill when adding a new NodeBridge handler to src/nodeBridge.ts, including updating types in src/nodeBridge.types.ts and optionally testing with scripts/test-nodebridge.ts

13

mcp-builder

anthropics

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

136215

Search skills

Search the agent skills registry