How much context does a browser game bug need before another developer can actually reproduce it?

[Hirsohi]

How much context does a browser game bug need before another developer can actually reproduce it?

I've noticed that a lot of bugs in browser games are not difficult because the fix is technically complicated. They are difficult because the useful context disappears between the person who finds the problem and the person who investigates it.

"Player sometimes freezes after returning to the tab" sounds simple until you start asking questions. Which browser and version? Was the tab backgrounded? Did requestAnimationFrame stop normally? Was the game restoring from a saved state? Was there an active WebSocket connection? Was a service worker involved? What was the device pixel ratio? Was the asset cache warm or cold? Was the bug dependent on frame timing, input order, or a random seed?

By the time all of this gets reconstructed, debugging the actual problem can be the easy part.

For HTML5 games I've started thinking that a good technical handoff should probably contain something closer to a small reproducible environment than a normal ticket.

Something like:

Build:
commit / build hash

Environment:
browser + version
OS
device
viewport
devicePixelRatio

Initial state:
save version
level / scene
random seed
relevant feature flags

Reproduction:
1. load level
2. move to X
3. background tab for 8-10 seconds
4. return
5. press input before next animation frame

Observed:
player position updates
camera does not

Expected invariant:
camera transform should match player transform after resume

Evidence:
console output
network errors
performance trace
relevant source files

For deterministic systems you could take this further and attach an input trace plus the seed, then replay the exact sequence. For timing bugs it gets harder because reproducing the environment may matter more than reproducing the inputs.

The idea clicked for me while looking at Wagglet, where the work being handed over carries context, constraints, and an expected outcome with it. It made me wonder whether we underspecify debugging tasks in game development, especially browser games where the runtime environment has so many variables.

I'm curious how other people handle this.

If another developer has to investigate one of your bugs without talking to you first, what information do you consider the minimum useful package?

Has anyone here gone as far as recording input events, random seeds, frame timings, visibility changes, network state, or game state snapshots so a bug can be replayed rather than manually reproduced?

I'd especially be interested in approaches that have worked well for Phaser, Pixi, Babylon, or plain Canvas/WebGL projects.