agentleFS
kensauruskensaurus/cursor-kenji/skills/backend-realtime/SKILL.md

backend-realtime

Skill9 starsChanged 42 days agoMITView on GitHub

Implement real-time features with WebSockets, Supabase Realtime, or Server-Sent Events. Use when "real-time", "live updates", "WebSocket", "notifications", "chat", "presence", "collaborative", "live data", or "instant sync".

  • Reads credentials

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read the roster from presenceState() on the sync event.

    line 89
  2. Very specific:

    Use for cursors and typing.

    line 98
  3. Very specific:

    Keep useQuery as the data source.

    line 104
  4. Very specific:

    Add the sender to typingUsers and drop them after 3 s of silence.

    line 141
  5. Distinctive:

    After implementing real-time features:

    line 147

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Real-time Features Skill

Degree of freedom: MIXED. Which transport and event [HIGH freedom]; existing-subscription probes and unmount cleanup [LOW freedom — run exactly].

How to reason

  1. Observe — existing channels, sockets, hooks, cleanup
  2. Interpret — new subscription vs reuse vs transport mismatch
  3. Classify — postgres-changes / presence / broadcast / SSE / optimistic
  4. Severity — leaked or duplicate subscription outranks a missing typing indicator

Worked example

Observe: chat mounts useRealtimeMessages twice; no untrack on leave; package.json already has @supabase/supabase-js. Interpret: duplicate INSERT listeners; leftover presence. Classify: one channel hook with unmount cleanup; reuse Supabase Realtime — do not add Socket.io. Verify: one postgres_changes subscription; removeChannel on unmount; no second socket lib.

Self-critique before reporting

  • Existing first — grepped channels/sockets before adding a library
  • Cleanup — every subscribe has unmount removeChannel / close
  • One listener — no duplicate subscription for the same event
  • Right owner — FE↔BE payload mismatch → debug-fe-be-integration; queue/job instead of live push → backend-patterns

Implement live, collaborative features using WebSockets, Supabase Realtime, and Server-Sent Events.

Code examples assume a Next.js App Router + Supabase stack (@/lib/supabase/client, 'use client'). Adapt import paths and row types to the detected stack.

Check existing first [LOW freedom — run exactly]

Before implementing ANY real-time feature, verify:

  1. Check for existing real-time setup:
cat package.json | grep -i "socket\|realtime\|pusher\|ably"
rg "supabase.*channel|useSubscription|WebSocket" --type ts --type tsx
  1. Check for existing patterns:
rg "on\('INSERT'\|on\('UPDATE'\|subscribe\(" --type ts
ls -la src/hooks/use*Realtime* src/lib/realtime* 2>/dev/null
  1. Check Supabase config:
rg "createClient|supabaseUrl" --type ts -l
grep -oh "SUPABASE_[A-Z_]*" .env* 2>/dev/null | sort -u   # names only — never print values

Why: Real-time connections are stateful. Don't create duplicate subscriptions.

Supabase Realtime [HIGH freedom]

Subscribe to Database Changes

The canonical useRealtimeMessages hook (initial fetch + INSERT/DELETE subscription with unmount cleanup) is in references/patterns.md.

useEffect(() => {
  const channel = supabase
    .channel(`room:${roomId}`)
    .on('postgres_changes', { event: 'INSERT', schema: 'public', table: 'messages', filter: `room_id=eq.${roomId}` },
        (payload) => setMessages((prev) => [...prev, payload.new as Message]))
    .subscribe()
  return () => { supabase.removeChannel(channel) }
}, [roomId, supabase])

Presence (Online Users)

  • One channel per room (presence:${roomId}); channel.track(currentUser) only after status === 'SUBSCRIBED'.
  • Read the roster from presenceState() on the sync event; join/leave are for side effects only.
  • Cleanup is channel.untrack() then supabase.removeChannel(channel).

Full usePresence hook: references/patterns.md §Presence.

Broadcast (Custom Events)

  • Hold the channel in a useRef; register .on('broadcast', { event }) handlers once, then .subscribe().
  • Send with channel.send({ type: 'broadcast', event, payload }); broadcast is ephemeral, never a source of truth.
  • Use for cursors and typing; use postgres_changes for anything that must persist.

Full useBroadcast hook and collaborative-cursor usage: references/patterns.md §Broadcast.

TanStack Query + Real-time [HIGH freedom]

  • Keep useQuery as the data source; the realtime channel only calls queryClient.invalidateQueries({ queryKey }).
  • Subscribe to postgres_changes with event: '*' on the table (optional filter); remove the channel on unmount.
  • One hook, useRealtimeQuery(queryKey, queryFn, table, filter?), so every list gets the same behaviour.

Full useRealtimeQuery hook: references/patterns.md §TanStack Query + Real-time.

Optimistic Updates [HIGH freedom]

  • useOptimistic(initial, reducer) plus useTransition; add the optimistic row with pending: true and a crypto.randomUUID() id, then await the Server Action.
  • Render pending rows at reduced opacity; the realtime INSERT or the action result replaces them.
  • Disable the submit button while isPending.
const [optimisticMessages, addOptimisticMessage] = useOptimistic(
  initialMessages, (state, msg: Message) => [...state, msg]
)
startTransition(async () => {
  addOptimisticMessage({ id: crypto.randomUUID(), content, pending: true, /* ... */ })
  await addMessage(content) // Server Action
})

Full Chat component: references/patterns.md §Optimistic Updates.

Server-Sent Events (SSE) [HIGH freedom]

  • Route handler returns a ReadableStream; write data: ${JSON.stringify(x)}\n\n frames with TextEncoder.
  • Headers: text/event-stream, Cache-Control: no-cache, Connection: keep-alive; heartbeat every 30 s.
  • Clean up on request.signal abort (clear timers, controller.close()).
  • Client: new EventSource(url); close in the effect cleanup and on onerror, then reconnect with backoff.
  • Pick SSE for one-way server push without Supabase; pick Realtime when the data already lives in Postgres.

Full route handler and useSSE hook: references/patterns.md §Server-Sent Events.

Typing Indicators [HIGH freedom]

  • Broadcast { event: 'typing', payload: { userId } } on keystroke; ignore your own id on receipt.
  • Add the sender to typingUsers and drop them after 3 s of silence; debounce sends with a timeoutRef.

Full useTypingIndicator hook: references/patterns.md §Typing Indicators.

Validation [LOW freedom — do not skip]

After implementing real-time features:

  1. Connection handling → Reconnects on disconnect, shows status
  2. Error handling → Graceful degradation when real-time unavailable
  3. Memory leaks → All subscriptions cleaned up on unmount
  4. Duplicate subscriptions → No multiple listeners for same event
  5. Optimistic updates → UI feels instant, handles conflicts
  6. Mobile → Works on spotty connections, battery efficient

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.