Daily Digest

Daily Digest

atlcli digest shows, in one view, what changed in Jira and Confluence since your last look and what is waiting on you. It runs a fixed, small set of queries (about 20 requests with the defaults), so it is fast enough for a shell profile or a cron job.

Prerequisites

• An authenticated profile with Jira and Confluence access (atlcli auth login)
• Jira permission: Browse Projects for the issues you work on
• Confluence permission: View for the spaces you work in
• For the sprint section: a scrum board, set with --board or global.board (kanban boards have no sprints and are skipped)
• For the project overview: Jira project keys, set with --project or digest.projects
• For --summarize: ANTHROPIC_API_KEY in the environment

Steps

1. Run the digest:

   atlcli digest

   The first run covers the last 24 hours. Each later run starts where the previous one ended (see The last-run marker).

2. Look back over a fixed window instead, without moving the marker:

   atlcli digest --since 7d

3. Add an overview of whole projects, whoever worked on them:

   atlcli digest --project GROW,VDS --since 2d

   To get the overview on every run, set digest.projects once (see Project overview).

4. Paste the Markdown version into a standup note:

   atlcli digest --markdown > standup.md

What the digest shows

Product	Section	What it lists
Jira	Waiting on you	Unresolved issues assigned to you that are not in a done status, by priority
Jira	Changed	Issues you are assigned to, reported or watch that changed in the window, each with its changes (status: To Do → In Progress by Ann · 3h ago)
Jira	Sprint	Each active sprint of the board: progress bar, to do / in progress / done counts, your open issues, days left
Jira project	Completed · Created · Changed	Per project from --project or digest.projects: issues moved into a done status, issues created, and issues changed in the window, by anyone, with assignee and change lines
Jira	Audit errors	Error-level findings of the audit rules for those projects; only when the audit feature is enabled
Jira	Team	Per person in those projects: open issues, in progress, overdue, no update for N days, and the oldest issue in progress. Not limited to the window
Confluence	Mentions	Pages and comments that mention you
Confluence	Comments on your pages	New comments on your 50 most recently modified pages
Confluence	Open inline comments	Unresolved inline comments on your 10 most recently modified pages (--max-pages), oldest first. These are not limited to the window: they wait on you.
Confluence	Your pages edited	Your pages that someone else edited
Confluence	Watched pages changed	Pages and blog posts you watch that someone else edited
Confluence	Recently changed in space	Pages changed in --space (or global.space); only when a space is set

A page that matches several sections appears once, in the first one in this order: mentions, your pages, watched, recent. In --json output its entry lists the other sections in alsoIn.

Backlog reordering (Rank) is not a change line; an issue whose only change is its rank is not listed.

Your own edits and comments are left out. Pass --include-own to keep them; your own Jira changes are then marked (you).

Empty sections are not printed. If nothing changed, the digest says Nothing new since <window>.

Options

Option	Type	Default	Description
--since	lastrun, <n>m, <n>h, <n>d, <n>w, today, yesterday, ISO date or datetime	lastrun	Start of the window. m means minutes here (in jira worklog it means months). today and yesterday start at local midnight. At most 30 days back.
--only	jira or confluence	both	Run one product only
--space	space key(s), comma-separated	global.space	Enables “Recently changed in space”
--board	board id	global.board	Enables the sprint section (scrum boards only)
--project	project key(s), comma-separated	digest.projects	Enables the project overview
--status-map	"<status>=todo|wip|done,…"	flow.statusMap	Override status categories, see Status mapping
--limit	integer ≥ 1	10	Items per section; a cut section is listed under warnings
--max-pages	integer ≥ 1	10	Your pages checked for open inline comments
--include-own	flag	off	Keep your own edits and comments
--no-mark	flag	off	Never move the last-run marker
--markdown	flag	off	Markdown output with ##/### headings and links
--json	flag	off	JSON output (schemaVersion: "1"); every section key is present, [] when empty, and the queries are included
--summarize	flag	off	Add a short summary from Anthropic, see Summaries
--verbose	flag	off	Print every JQL/CQL query, the request count and the duration to stderr

Project overview

The personal sections only show what concerns you. The project overview shows everything that happened in whole projects: per project, the issues completed (moved into a done status in the window), created and changed, by anyone, including you.

Set the projects once in ~/.atlcli/config.json, globally or per profile. A profile’s list replaces the global one, and --project replaces both:

{
  "digest": { "projects": ["GROW"] },
  "profiles": {
    "work": { "digest": { "projects": ["GROW", "VDS", "ADMIN"] } }
  }
}

Option	Type	Default	Constraints
digest.projects	string[]	[]	Jira project keys

The overview reads up to 200 changed issues across all projects in one search, plus one bulk changelog request. Each list is cut at --limit.

“Done” follows the Jira status category (To Do, In Progress, Done), not status names. If a workflow puts a done status in another category, map it with flow.statusMap or --status-map, e.g. "Fertig=done". The map also applies to “Waiting on you” and the sprint counts.

Audit errors

With the audit feature enabled (atlcli flag set audit true --global), the digest also runs the audit rules for its projects and lists what is still at error after the baseline. It uses the same baseline as atlcli audit jira and the same rules file, except one in the current directory’s .atlcli/: the digest reads only the profile and global rules files. It runs only the rules that can reach error, and leaves out flow/stuck-wip and flow/status-aging, which read the done issues and their history, unless the rule sets digest: true.

The section reads up to 300 open issues across the projects; beyond that, a warning says the list is a lower bound. With the default rules, the section costs a few requests: one search for open issues, one for resolved issues outside Done, one bulk changelog request, and one field list for the Flagged field.

Jira
  Audit errors (2)
    GROW-60  [In Progress]  HR marketing — hygiene/overdue: Due 2025-12-31, 294 d overdue.
    GROW-1343  [In Progress]  Meetups — hygiene/stale: Not updated for 157 d.

Team

With projects (--project or digest.projects), the digest lists everyone with open issues in them, busiest first, and the unassigned issues last:

Jira
  Team (3)
    Thomas Schindler  open 4 · in progress 4 · no update 30d+ 4 · oldest in progress 469d
    Roman Binshtok  open 2 · in progress 0 · no update 30d+ 2
    (unassigned)  open 56 · in progress 7 · no update 30d+ 55 · oldest in progress 486d

Figure	Meaning
open	Issues not in a done status, after flow.statusMap
in progress	In an In Progress status, after the map
overdue	Due date before today
no update Nd+	Not updated for N days or more: the first value of audit.jira.staleDays (default 30)
oldest in progress	Days since the longest-waiting issue in progress was last updated

The section costs one search (up to 300 open issues across the projects; beyond that, a warning says the counts are a lower bound). --summarize sends these names and counts too.

The last-run marker

The marker lives in ~/.atlcli/profiles/<profile>/digest-state.json, with one entry per product:

{
  "schemaVersion": 2,
  "jira": { "lastRunAt": "2026-10-22T07:58:12.311Z" },
  "confluence": { "lastRunAt": "2026-10-22T07:58:12.311Z" }
}

• A product’s marker moves to the start of the run, so changes made while the digest ran are not lost.
• It moves only when --since was not given (or was lastrun), none of that product’s sections failed and --no-mark was not passed. Peeking with --since 7d never moves it.
• Each product starts from its own marker. If Confluence was down yesterday, today’s run catches up on Confluence while Jira stays current.
• The window starts 5 minutes before the marker, because search indexes trail writes. Items in that overlap can show up twice across two runs.

Summaries

--summarize sends the rendered Markdown digest to Anthropic in one model call and prints a short summary above the digest (in --json, in the summary field).

What is sent

Only the digest itself: issue keys, page and issue titles, statuses, people’s names and section names. No page bodies, issue descriptions or comment text. Before the call, the CLI prints Sending digest (N items: …) to Anthropic. to stderr.

• The key is read from ANTHROPIC_API_KEY only. Flags such as --api-key are refused.
• Without a key, or if the call fails, the digest is still printed with a warning, and the exit code stays 0.
• The model treats titles as data, never as instructions.

Examples

Minimal:

atlcli digest

Jira only, last week, with the sprint of board 42:

atlcli digest --only jira --since 7d --board 42

A standup note every weekday at 8:45, with a summary:

# crontab -e
45 8 * * 1-5  ANTHROPIC_API_KEY=... atlcli digest --markdown --summarize > ~/notes/standup.md 2>/dev/null

Debug which queries run:

atlcli digest --since 2h --verbose --json > /dev/null

Exit codes and partial failures

Each section is fetched on its own. A failing section is listed under Warnings and the rest is still printed.

Situation	Exit code	Marker
Everything succeeded	0	Moves for both products
Some sections failed	0	Moves only for products without a failed section
Every section failed	1	Unchanged

Troubleshooting

A fresh mention or comment is missing. Confluence search can trail writes by up to a minute. The next lastrun digest picks it up, because its window overlaps the previous one by 5 minutes.

“Changed” lists issues without any change lines. On Jira Data Center there is no bulk changelog API, so the digest lists the issues without change details.

The marker did not move. A section of that product failed (see the warnings), or the run used --since or --no-mark.

truncated: jira.waiting under warnings. More items matched than --limit. Raise --limit, or use atlcli jira my for the full list.

Finished issues show up in “Waiting on you”, or “Completed” stays empty. The workflow puts a done status into the To Do or In Progress category. Map it: --status-map "Fertig=done" or flow.statusMap. Better still, move the status to the Done category in the Jira workflow.

truncated: jira.projects under warnings. More than 200 issues changed in the projects during the window. Narrow --since or the project list.

The digest reaches back only 30 days. That is the upper bound, to keep queries cheap after a long break. A warning says so.

Related topics

• Jira search and jira my
• Confluence search: wiki my and wiki recent
• Configuration: defaults and profiles
• Jira Audit and Audit Rules
• Research, the agentic alternative for open questions
• Automation Scripts, for changes across many issues and pages