Bug Report Template (Free, Copy-Paste) for Web, Mobile & API

This is the transactional version of a bug-report guide: you came for the template, so the templates are first, and you can copy any of them in one click. If you want the reasoning behind why these fields matter — what good looks like and why tickets bounce — read the companion piece, how to file a bug report developers actually want to fix. This page stays focused on the artifact.

One generic form does not fit web, mobile, and API bugs. A web bug needs the viewport and console; a mobile bug needs the device build and network type; an API bug needs the exact request and the full response. Forcing one shape onto all three is the quiet reason reports come back marked need more info.

This is not opinion. In Assessing the Quality of the Steps to Reproduce in Bug Reports (ESEC/FSE 2019), Chaparro and colleagues built a tool called EULER to auto-grade reproduction-step quality — it correctly identified about 98% of the reproduction steps that were present, but it also flagged roughly 58% of the steps that were missing. The tool exists because incomplete steps are common enough to be the headline failure mode worth automating against.

The cost shows up downstream. A large-scale study of non-reproducible bugs (Rahman & Roy, 2022) analyzed 576 non-reproducible reports from Firefox and Eclipse, isolated 11 recurring causes, and ran a user study with 13 professional developers — who responded by either closing the bug or asking for more information. That back-and-forth is the exact tax a complete template removes.

Use this for anything that happens in a browser: a broken layout, a failed form submit, a console error. The fields beyond steps and expected-vs-actual are the ones a capture tool should fill in automatically — keep them in the template anyway, so a manual reporter has a slot for each.

Mobile adds variables a web template ignores: the exact device model, the OS build number (not just "iOS"), the network type, and — critically — whether the bug reproduces on Wi-Fi versus cellular. Race conditions and timeouts often only appear on one. Capture the build number, because "latest" is not a version.

For an API bug, the report is the request and the response. A triager should be able to paste your curl command and watch the same failure. Give the exact method, endpoint, headers (redact secrets), and body, then the full response with its status code — and the environment or base URL, because staging and prod fail differently.

A Markdown template is a suggestion; a reporter can delete the headings and type one line. A GitHub issue form is enforced. Commit the YAML below to .github/ISSUE_TEMPLATE/bug_report.yml and GitHub renders a structured form with required textareas, a browser dropdown, and a type: bug label. With required: true on the steps field, a reporter physically cannot submit without it — which is exactly the field they'd otherwise skip.

GitHub's issue-form schema supports more field types than this — input, textarea (with code rendering), dropdown, checkboxes, file upload, and a type: bug designation — so you can grow the form to match the web, mobile, or API spine above. If your team already lives in GitHub Issues, this is the cheapest way to raise report quality across the board.

Templates fix the fields a human must write. They do nothing for the fields a human forgets — and environment is the second-most-skipped field for a reason. A capture tool inverts the problem: instead of asking the reporter to remember their browser, OS, viewport, build SHA, console errors, and failing network calls, it records them automatically the moment the bug happens. The reporter is left with the one field that genuinely needs a human: a clear sentence and the steps.

With the BugMojo browser extension, a teammate highlights the broken element, types one line, and submits. Attached automatically: the last ~30 seconds of DOM mutations as an rrweb session replay, every console message in that window, every network request (status, payload, headers — PII redacted in the browser first), and the full environment block. The session replay means the steps to reproduce are no longer prose to be argued over — they are a recording you press play on.

Every bug-report template you'll find — Atlassian's, GitHub's, and the rest — optimizes for a human triager. None of them ask the next question: can an AI coding agent read this bug and start fixing it? A structured template gets you halfway, because consistent field names let an agent parse steps, environment, and expected-vs-actual. The missing half is access — the agent still needs a way to fetch the evidence.

Be honest about the trade. A Markdown or YAML template is free, portable, and works in any tracker; that's a real advantage and BugMojo does not replace it. Where the templates stop is the machine handoff.

BugMojo closes the access gap with an MCP server that exposes captured bugs — session replay, console, network, and environment — directly to agents like Claude Code or Cursor. The agent reads the same evidence a human triager would, instead of guessing from a one-line title. That's the row no competitor template page has, and it's the difference between a report a person re-traces by hand and one an agent can act on.

• Pick the right template. Web, mobile, or API — paste it into Linear, Jira, or GitHub and set it as the default issue template.
• Enforce the one field that gets skipped. If you're on GitHub, ship the bug_report.yml form with required: true on steps to reproduce.
• Autocapture the rest. Environment, console, and network are fields a tool should fill in — a reporter cannot forget what they never had to type.
• Make it agent-readable. If you want Claude Code or Cursor to act on bugs, the report has to be both structured and fetchable.