Bug Report Template — Repro Steps Devs Actually Need

“It crashed.” Black screenshot. No OS. No build. No steps. That ticket sits in triage for two days while someone DMs “can you repro?” A usable bug report is a gift to future-you and the person who ships the fix — especially for game crashes where “sometimes when fighting” wastes a sprint.

The minimum viable report

Field	Why it matters
Title	Searchable one-liner (Crash on save when inventory full)
Build / version	“Latest” is not a version — use commit, build number, or store version
Platform	Windows 11, iOS 18.1, Chrome 126, Steam Deck…
Hardware extras	GPU, controller, headset — for games especially
Repro rate	1/1, 3/10, once-only
Steps	Numbered, start from a known state
Expected / Actual	What should happen vs what happened
Evidence	Log snippet, stack, clip, save file

Draft structure with a bug report template, then paste into GitHub/Jira/Discord and attach evidence. Fill environment before you write the novel.

Repro steps that do not waste time

Bad: “Open the game and mess around until it dies.”

Better:

1. New game → difficulty Normal
2. Reach tutorial chest (first room)
3. Fill inventory to 20/20
4. Open inventory → drag potion onto hotbar slot 1
5. Actual: freeze 3s → desktop
   Expected: potion equips

One action per line. Name UI labels exactly as shown. Note timing for race bugs (“within 1s of landing”). If a save file or seed is required, say so in step 0.

Games vs web vs mobile

Games: graphics preset, fullscreen vs windowed, VSync, overlays (Discord/NVIDIA/Steam), mods, save/seed, multiplayer region. Attach Player.log / crash dumps when the engine provides them.

Web: browser + version, account role/feature flags, exact URL (redact secrets), console errors, Network status codes, whether hard refresh changes anything.

Mobile: device model, OS build, install source (TestFlight, Play internal track, store), storage pressure, backgrounded vs foreground crash.

Evidence without dumping secrets

Good	Risky
Last 50 log lines, IDs scrubbed	Full .env / auth tokens
10–30s clip of the freeze	Hour-long desktop stream with Slack open
AbortError stack	Customer PII / chat contents in screenshots
Redacted HAR of the failing call	Session cookies in HAR

If legal or support needs the full dump, share via a private channel with retention limits — not a public Discord.

Severity without ALL CAPS

Pair impact with scope — not just “URGENT!!!”:

• Blocker — cannot purchase / soft-locks new users / data loss
• Major — frequent crash on a common path
• Minor — cosmetic, rare, workaround exists

Useful labels: crash, softlock, regression, platform-pc, needs-repro. Three severity levels beat twenty synonyms. Note if this is a regression from build N−1 — that halves investigation time.

Flaky and once-only crashes

If repro is 1/20, say so honestly and list suspects (overlays, low disk, thermal throttling, specific locale). Capture:

• Exact time (UTC) for log correlation
• Whether other apps crashed
• Whether it happened after sleep/resume

Do not force a “100% repro” claim you cannot defend. Engineers prioritize differently when they know it is intermittent.

Player and support intake

Rewrite messy player emails into the template before eng pickup — five minutes of cleanup saves fifty of clarification. Support macros that ask for build + platform + steps up front pay for themselves. For Discord communities, pin the skeleton and reject empty “game broken” posts gently with a link.

Copy-paste skeleton

### Summary

### Environment
- App / build:
- OS / device:
- Repro rate:

### Steps
1.
2.

### Expected

### Actual

### Attachments

Triage hygiene for teams

• Duplicate search by title keywords before filing
• One bug per ticket (do not bundle three crashes)
• Update the ticket when you find a shorter repro
• Close with the fixing PR / patch notes build number

Unstructured reports create ping-pong. Structured ones get labeled and fixed. Next crash — fill the fields once, attach evidence, and skip the two-day “what build?” loop.