Measure delivery

Check that your agent can reach Jira, then ask for the numbers your team cares about. The agent only reads, so your tickets never change.

Check Jira access

Confirm your credentials work before you ask for numbers:

Run the Jira access check and tell me whether my credentials work.

The agent runs the jira skill’s check script. If it fails, set up your credentials first with the credential-setup skill, because the metrics tool never reads them itself.

Measure a project

Ask in team language:

What's our cycle time this quarter for PROJ?

The agent runs the flow-metrics skill and returns cycle time, lead time, throughput, work in progress, and the rest of the Flow Framework and DORA set as JSON for the last 90 days. Name a different window, or a filter such as “only issues labelled ai-assisted”, to narrow it.

Split out a group

To see how one group of work compares with the rest, name the group:

Compare issues labelled ai-assisted against everything else in PROJ this quarter.

The agent adds a breakdown with the group on one side and the rest on the other.

Roll up a program

For a program or portfolio, give its ID instead of a project:

Measure delivery for Jira Align program 42 from February to April.

The agent joins the program’s issues through jira-align, which must be installed and signed in.

Compare two periods

Ask for each period, then for the comparison:

Measure PROJ for Q1 2024 and Q4 2025, then write a report comparing them.

The agent runs flow-metrics for each window and pairs the results with ai-adoption-report into a Markdown report and a JSON file. Both periods must cover the same scope, and the earlier one must end before the later one starts.

Results are cached under .context/flow-metrics/cache/, so ask for a fresh run when your Jira data has changed. To measure the build loop itself, export loop telemetry from core.

Next, report AI adoption as a delivery lead. The atlassian skills reference lists every flag.