Daily nudge for stalled Jira sprint issues

Build an agent workflow that runs on a cron trigger every weekday at 9am in my local timezone and nudges stalled work in my active Jira sprint.

Step 1. Use Jira Search Issues (JQL) with a query like: sprint in openSprints() AND status not in (Done, Closed, Resolved) AND updated < -3d ORDER BY updated ASC. Page through all results and collect the issue key, summary, status, assignee (account id, display name, and email if available), updated timestamp, and the description.

Step 2. For each stale issue, call Jira Get Comments and pull the most recent comments (last 5 is plenty). Also keep the description from step 1. Before doing anything else, check the comments for a marker that says this workflow has already commented in the last 24 hours (for example, look for an HTML comment marker like <!-- generalinput-sprint-nudge --> in the comment body, plus a created timestamp within the last 24 hours). If found, skip this issue entirely.

Step 3. For each remaining stale issue, have the agent read the description and recent comments and write a short diagnosis (two to four sentences): what looks blocked, what the most likely next step is, and any specific question to ask the assignee. Keep it concrete and grounded in the actual content of the issue, not generic.

Step 4. Call Jira Add Comment on each stale issue with a friendly message addressed to the assignee, embedding the diagnosis and a clear ask for a status update or next step. Include the hidden marker <!-- generalinput-sprint-nudge --> at the end of the comment body so the next run can detect it. Use Atlassian Document Format for the comment body.

Step 5. After commenting on all of them, post ONE rolled up message to the team Slack channel using Slack Bot Send a Message. Group issues by assignee, @mention each assignee using their Slack user (look them up by email if needed), and under each name list the stale issues as bullets with the issue key as a link to the Jira issue, the summary, how many days since last update, and a one line version of the diagnosis. Open the message with a brief header like 'Sprint health check, {date}' and a one line summary of total stale issues. If there are zero stale issues, post a short 'No stalled issues today, nice work' message instead of skipping silently.

Inputs the workflow should accept when I configure it: the Slack channel id or name to post the digest to, an optional Jira project key or board id to scope the JQL (default is all open sprints I can see), and the staleness threshold in days (default 3).

Be polite and human in the Jira comment tone, not robotic. Never escalate, never @channel in Slack, and never repeat a nudge on the same issue within 24 hours.