Multiplayer
@multiplayer.app
55 followers 180 following 560 posts
Full stack session recording. End-to-end visibility in a single click. Capture once, use for debugging, testing, support, feature development ... Try it for free: multiplayer.app
Posts Media Videos Starter Packs
Pinned
multiplayer.app
Full stack session recordings: record. code. fix. repeat.
multiplayer.app
Also… Remote recording is now in beta. Silently enable recording of user experiences, ensuring nothing gets lost in translation from vague bug reports.
multiplayer.app
You can also enable continuous recording which keeps a lightweight rolling buffer that auto-saves frontend + backend exceptions and errors the moment they appear.

No repro steps. No missed bugs.
multiplayer.app
Yes, you can absolutely capture on-demand session replays.

Manually start / stop to capture the exact reproduction steps

BUT…
multiplayer.app
Most session replay tools force a tradeoff:

🎥 Record on demand → you miss what happens outside the window.
📼 Record everything → you drown in noise and costs.

We took a different approach 🧵
multiplayer.app
Sometimes the frontend data isn’t enough.

Sometimes (okay, always) you also want to know what happened in the backend.
multiplayer.app
“Is this the real bug? Or it’s just fantasy?” 👈 Strong start to my chat with Freddie Mercury about our latest bug.

Luckily, Multiplayer caught the whole performance: frontend, backend, and the error’s big solo.
multiplayer.app
The Multiplayer API key authenticates your application when sending session recording data to Multiplayer.

The API key ensures that session recordings, traces, logs, and other data are securely tied to your account.
multiplayer.app
👀 Did you know that with Multiplayer you can:

👉 Use a single API authentication key for both frontend and backend data,

OR

👉  Generate separate API authentication keys for frontend and backend if you want stricter separation or different access scopes
multiplayer.app
(3) Notebook: add the API requests that reproduce the flow + assertions for expected outcomes.

(4) AI via MCP: “Generate a patch to satisfy this recording’s notes”

(5) Run the notebook to validate the fix; attach results to the PR.

(6) Keep the recording + notebook as living documentation.
multiplayer.app
Here’s an example of a concrete SDD workflow with Multiplayer:

(1) Capture a session where the bug or desired behavior appears.

(2) Annotate the exact UI state, API call, and trace span with requirements.
multiplayer.app
Specs don’t have to be static documents anymore.

With Multiplayer, they can be:

✅ Captured from real behavior
✅ Annotated with requirements
✅ Run as executable notebooks
✅ Fed directly into AI tools for accurate fixes & features
multiplayer.app
A full stack session recording + annotations fits perfectly with this shift:

✔️ Frontend screens + backend traces
✔️ Requests/responses + headers
✔️ Notes, sketches, requirements

Executable with Notebooks, shareable with your team, and consumable by AI through our MCP server.
multiplayer.app
So why the renewed attention now?

Because AI needs structured context.

A good spec → fewer hallucinations, more accurate code, tests, and features.

Without it, copilots and agents are left guessing.
multiplayer.app
Spec-driven development flips the script:

‣ The spec is the source of truth
‣ It’s machine-readable and executable
‣ From one spec, you can generate docs, SDKs, tests, and implementations

Think OpenAPI, gRPC protos, GraphQL SDL.
multiplayer.app
Traditionally, specs were an afterthought.

Teams wrote them once, then let them drift while “the real work” happened in code.

Code became the source of truth. Specs became stale PDFs or wikis.
multiplayer.app
Spec-driven development is making a comeback.

Why? Because AI-assisted coding has changed the game.

Let’s break it down 👇
multiplayer.app
With zero sampling, you don’t rely on luck. You get:

✔️ Confidence that the trace you need is there

✔️ Precision debugging across the full stack

✔️ AI-ready context: feed your IDE/copilot full correlated data + annotations
multiplayer.app
You can also enrich your session recordings with request/response content and headers.

So your replays carry not only what happened, but how and why.
multiplayer.app
How to do it differently:

‣ We create a unique trace context for every session
‣ We override sampling at root so all spans are kept
‣ As requests flow across services, every span preserves “sampled = true”
‣ We correlate all that backend data with frontend UI actions in a unified timeline
multiplayer.app
Replay tools usually stop at the browser.

If they capture backend data at all, they inherit the sampling behavior of your APM tool or require fragile, manual setups.

The result: fragmented context and painful debugging across tools.
multiplayer.app
🔍 The sampling problem

APM tools ingest everything, but to control cost they sample aggressively.

That’s okay for system health and trend monitoring, but when you’re debugging a user issue, you may miss the exact trace you needed.
multiplayer.app
Most tools sample your telemetry, leaving gaps when you desperately need precise data.

Here’s how 👇 to get * unsampled * backend traces, and then stitch them to your frontend sessions.
multiplayer.app
For us that’s meant designing MCP tools that help developers fix bugs and build features faster by feeding AI copilots the right context: from full-stack traces to annotated session data.

If you’re experimenting with MCP, think small, scope tightly, and expand based on how users actually work.