Blog

Notes from building the bug hand-off layer

Practical writing on bug reporting, browser debugging, and AI-agent workflows. No filler posts.

AI debugging2026-08-01 6 min

How to report bugs to Cursor so it fixes them on the first try

Cursor is brilliant at code and blind to your browser. The difference between a five-round guessing session and a first-try fix is what you paste into that first message. Here's what works — manually, and with one click.

Read
React2026-08-01 7 min

Debugging React bugs with AI: why your agent guesses, and how to make it stop

React's most common bugs — undefined props, stale state, effects firing at the wrong time — are exactly the ones AI agents struggle to fix from a description alone. What each bug class needs as evidence, and how to capture it.

Read
QA2026-08-01 7 min

The QA workflow where your bug reports get fixed by AI (before standup)

QA captures the bug once, with everything attached. The developer's AI agent reads it and starts fixing. A practical workflow for QA teams working with AI-assisted developers — including Sentry mode, the arm-once-capture-all-day trick.

Read
AI debugging2026-07-31 7 min

How to give Claude Code full browser context for a bug (stop typing 300-word descriptions)

Claude Code can't see your browser. When a UI bug happens, you become its eyes — badly. What coding agents actually need to fix a browser bug, how to assemble it by hand, and how to capture all of it in one click.

Read
Playwright2026-07-31 6 min

Turn a bug report into a failing Playwright test — automatically

A bug report tells you something broke. A failing test proves when it's fixed. How TraceBug generates a runnable Playwright spec from a captured browser session — red while the bug exists, green after the fix — and why that changes AI-assisted debugging.

Read
Comparisons2026-07-30 9 min

7 Jam.dev alternatives for bug reporting in 2026 (an honest comparison)

Jam is a great tool — but it's not the only way to capture browser bugs. A frank tour of BetterBugs, Bird Eats Bug, Marker.io, BugHerd, Userback, Replay.io, and TraceBug: what each is actually for, and when to pick it.

Read
Comparisons2026-07-30 7 min

6 BetterBugs alternatives for bug capture in 2026 (picked by workflow, not features)

BetterBugs is a capable Jam-style capture tool — but the right alternative depends on where your bug reports end up: a team dashboard, a client's PM tool, or an AI coding agent. An honest tour of Jam, Bird Eats Bug, Marker.io, BugHerd, Userback, and TraceBug.

Read
Bug reporting2026-07-19 6 min

How to write a bug report developers will actually read (template included)

Most bug reports get bounced back with questions, not fixed. A working template, the five details that let a developer reproduce your bug on the first try, and the mistakes that get reports ignored.

Read
Under the hood2026-07-19 5 min

What is session replay? (It's not a screen recording)

Session replay records the DOM, not pixels — that's why a replay is a few hundred kilobytes, lines up with your console and network logs to the millisecond, and can mask passwords before they're ever stored.

Read
AI debugging2026-07-19 5 min

A practical MCP server example: let Claude read bug reports from your disk

A concrete Model Context Protocol example you can run in one command: a local, stdio-only MCP server that gives Claude Code, Cursor, or Windsurf read tools over exported bug reports — with a real transcript of the result.

Read
AI debugging2026-07-19 6 min

Stop pasting screenshots into Claude — give your AI agent the actual bug

AI coding agents are only as good as their evidence. How to hand Claude Code, Cursor, or Windsurf a complete bug report — replay, console, network, repro steps — through a local MCP server, in two minutes.

Read