Jira worklogs, filled in from time you never had to log

Logging work in Jira is a chore that survives every process improvement, because it asks you to remember on Friday what you were doing on Tuesday. BillNotch already knows: the desktop tracker recorded the window, the issue key was sitting in the title of the tab or the branch you had open, and the only thing left is for you to agree with it.

Jira Cloud · OAuth · worklogs out, issues in

What actually crosses between the two systems?

How does a tracked hour find its issue?

By reading the text the capture already produced. An issue key has a shape — PROJ-482, uppercase letters, a hyphen, digits — and that shape turns up constantly in the things you actually have open: the Jira tab itself, a pull request title, a branch name in your editor’s window bar, a Figma frame someone named after the ticket.

BillNotch pulls every key it finds in a stretch of activity, looks each one up in its local issue index, and takes the first one that belongs to a Jira project you have mapped to a BillNotch project. That match is strong evidence, so it arrives on the Review screen as a high-confidence suggestion — but a rule you taught the categorizer yourself still outranks it, because your correction is worth more than an inferred key.

The mapping is the part you have to do by hand, once. An unmapped Jira project matches nothing, on purpose: a key like OPS-1 passing through a support tab should not quietly redirect a billable afternoon.

How you connect it

In Settings → Integrations, press Connect on the Jira row. Atlassian asks you to approve access to Jira work and users on your site, and returns you to Settings. Import your projects, map the ones you bill against, and turn on Sync worklogs to Jira if you want time written back — that switch is off until you set it, so connecting alone never writes anything to your tracker.

One Atlassian site per connection: if your account can reach several, BillNotch binds to the first and stays there. Connecting requires the integrations.manage permission, which owners and admins hold and ordinary members do not.

What a worklog contains

A start timestamp and a number of seconds. There is no comment on the request, and that is an invariant rather than a default someone might loosen later: window titles are the most sensitive thing an automatic tracker holds, and a Jira worklog is visible to everyone who can see the issue. Entries under a minute and entries with no project are skipped entirely.

Each write is claimed against a link row keyed to the entry, so the same hour cannot be logged twice however many times the sync runs. A failed write is retried a handful of times and then given up on and marked skipped, rather than sitting in a queue forever attempting to reach an issue somebody deleted.

When does the sync run?

Issues refresh in the background when you open the Review screen, and on demand when you press Sync now. Each run picks up only what changed since the last one, and repeated runs inside a quarter of an hour are ignored, so a busy team does not hammer the Jira API. There is no scheduler: if nobody opens BillNotch, nothing syncs, which is the honest shape of a product that does not want to be running jobs against your tracker at three in the morning.

Worklogs go the other way at a single moment — when you confirm entries on Review. Nothing the tracker captured reaches Jira before a person has agreed with it.

Why issue summaries never reach an AI model

Imported issue summaries do help categorization: they are keywords, and keywords are exactly what a deterministic matcher wants. But they are used only by the matcher that runs on BillNotch’s own servers. When the optional AI categorizer is enabled, the prompt it builds deliberately excludes them — your ticket titles are your company’s roadmap, and they do not get shipped to a model provider to save someone a few lines of code.

What it will not do

It will not move an issue through your workflow, comment on it, create it, assign it or read its status. It will not update or delete a worklog it already wrote, so editing an entry in BillNotch after the fact leaves Jira as it was. It does not import worklogs the other way, and it does not map projects for you.

And it is Jira alone. Linear, Asana, ClickUp, Monday, Trello, GitHub Issues and Notion are not built — if your tickets live in one of those, the tracker still captures the work and the categorizer still sorts it, you simply do not get worklogs written back.