Pipeline health, what's critical, what's stuck, who's overloaded, what came in overnight — delivered to your inbox, Slack channel, or Teams channel before standup. No dashboards to babysit. No saved-filter gymnastics. No servers to run.
The repo's README contains an interactive handoff prompt that turns any LLM into your personal setup assistant. It asks you which Jira board, what time of morning, which channels, then configures everything end-to-end. Setup time: ~20 minutes. No coding required.
Copies the full handoff prompt to your clipboard (≈ 15 KB of step-by-step setup instructions).
Claude Code, Codex, Cursor, Windsurf, Claude.ai, ChatGPT, Gemini — any of them, with your forked repo accessible.
It interviews you in plain English. Answer the questions ("which board?", "what time?", "Slack or email?"). Done.
Every PM, EM, and support lead does the same dance each morning. This replaces it.
Below is a real rendering — generated by the production code — with mock data. Every section header in your real digest is a click-through to the live Jira search.
Switch channels with the tabs above · hover the purple dots for what's configurable · click anything to see the JQL behind it. Company names and tickets are fabricated.
Opinionated where it should be. Configurable where you need it.
KPI counters, prioritized issue lists, status breakdowns, stuck-item detection, recent intake, custom-field groupings. Mix any combination.
Gmail (draft or send), generic SMTP, Slack (bot token or webhook), Microsoft Teams (incoming webhook). Use one or all.
Every section title and KPI counter is a click-through to a live Jira search. No screenshots, no stale snapshots — always current.
Write customer_account in JQL — the script swaps in customfield_XXXXX at runtime. Discover your IDs and alias them once.
Uses the Jira changelog to compute true dwell time — no false positives from unrelated edits. Threshold-configurable per section.
If the run dies or every channel errors, get pinged on Slack DM or email. The script doesn't fail silently.
GitHub Actions cron can run 1–3 hours late. Workflow accepts repository_dispatch from cron-job.org for second-precision delivery.
Optional. Each day's digest uploaded as a GitHub gist so Slack/Teams messages link to the full HTML when chat clients truncate.
One YAML config drives everything. Swap Jira instances, projects, recipients, channels by editing a single file. Zero code changes.
Each one is a building block. A typical digest uses four to eight.
Row of numbers — Open, New (7d), Closed (7d), Critical, whatever. Each count links to its underlying Jira search.
Table from any JQL. Pick columns, sort, limit. Best for "critical & high priority" and similar.
Same shape as list — names the intent. Pair with a time-window JQL like created >= -24h.
Count per active status. Shows zero-count statuses too, so stalls jump out.
Issues with no status change in N+ days. Uses Jira's changelog for accurate dwell time.
Group by assignee for bandwidth, by customer for impact, by component for hot-spot detection — works on any custom field.
Built for non-technical users. An LLM handles the wiring; you just answer questions.
One click on GitHub. The fork is yours — your secrets stay in your account, your config stays in your repo.
The README contains a self-contained interview prompt. Paste it into any LLM with file access and it walks you through Jira credentials, custom-field discovery, section selection, delivery channels, secret setup, and scheduling — answering plain-English questions like "which board?" and "what time of morning?"
Typical end-to-end time: 20–30 minutes, most of which is waiting on OAuth consent screens. Tomorrow morning, the first digest arrives.
Minimal dependencies. Boring tech, by design.
Fork it, paste the handoff prompt into your favorite LLM, and let it interview you. ~20 minutes from now you're done.