Metrics - Jira analytics for engineering teams

See how work really flows through your team - straight from Jira, in one command.

Metrics pulls your issues and turns them into clear charts plus a single shareable report:

• How fast are we? Lead time, cycle time, and a percentile view - "85% of our tickets finish within N days"
• When will it be done? A Monte Carlo forecast for your open backlog
• Where does work get stuck? Time per status, aging work-in-progress, and a cumulative flow diagram
• Who carries the load? Time in flight per person and how often tickets change hands
• How much do we ship? Weekly throughput and how often tickets bounce back to testing

Everything lands in output/ as PNG charts and one self-contained report.html you can send to anyone - no servers, no setup, nothing leaves your machine. The report is interactive: headline numbers with trend arrows, hover any dot for the ticket details, click it to open the issue in Jira, and a ready-made list of the longest-waiting tickets.

Real examples

Every chart below comes from real public projects - the Hibernate ORM and Apache Kafka issue trackers.

When will it be done?

Monte Carlo forecast. Half the simulations finish 142 open Hibernate issues by late September; 85% by 21 October - real dates, not story points.

Weekly throughput. How many issues get finished each week, with the overall trend.

How fast do tickets finish?

Cycle time, ticket by ticket. Every dot is a finished Hibernate issue - the last 4 weeks in blue, the slowest ticket named, and the "slow zone" beyond p85 tinted red.

Lead time. From the moment a ticket is created to the moment it's done.

Cycle time distribution. The same story as the scatterplot, as a simple histogram.

Where does work get stuck?

Cumulative flow. Two months of Hibernate: each band is a status, widening bands are bottlenecks - and notable events annotate themselves, like 59 issues leaving "Release pending" in one day.

Aging work-in-progress. Open Kafka issues by how long they've sat in their current status - anything above the dashed line has been waiting longer than 85% of past work ever did.

Median time per status. Where a ticket's calendar time actually goes.

Days in each status. Every status of the workflow at a glance, one histogram each.

How does the team work?

Who carries the work. Time in flight per assignee (names anonymized here) and how often tickets change hands.

Rework. How many times tickets came back to review - each return is work done twice.

Try it in one minute - no account needed

Point it at any public Jira, like Hibernate's:

uv sync
uv run python -m metrics --anonymous \
  --jira-server https://hibernate.atlassian.net \
  --jira-jql 'project = HHH AND created >= -60d' \
  --testing-statuses "In review, Waiting for review"

open output/report.html

Get started with your own Jira

uv run python -m metrics --jira-server https://your-jira --jira-token <token> --jira-jql 'project=MYPROJ'

Then open output/report.html.

• Jira Cloud: also pass --jira-email you@company.com together with an API token.
• Jira Server / Data Center: just a personal access token, no email needed.
• Public instance? --anonymous needs no credentials at all and figures out Cloud vs Server by itself.
• Your workflow ends differently? Tell it what "finished" means, e.g. --done-statuses "Resolved, Shipped" (default: done, completed, cancelled, closed, resolved).
• QA has its own name? Same for the rework metric, e.g. --testing-statuses "In review, QA" (default: testing).
• Curious how much time is real work vs waiting? Tell it where work happens, e.g. --active-statuses "In Progress, In Development" (default: in progress) - that powers the flow-efficiency number in the report.
• Prefer environment variables or a config file? uv run python -m metrics --help shows every option.

Tip: let your JQL include both finished and still-open issues - the forecast and aging charts need the open ones.

License

GPL - see LICENSE.