AI Agent Daily Standup

AI agents running overnight with amux post their completed tasks, blockers, and next steps to a shared kanban board — your morning review is a 5-minute board scan from your phone instead of a standup meeting. Updated July 2026.

Quick answer: Set up a morning report schedule — at 7am, each agent session summarizes yesterday's output to the board, and you review it over coffee. No meeting, no context-switching, no waiting for humans to post updates. Each standup post includes completed items, files changed, test results, PRs opened, blockers, and next planned action.

The Three-Part Morning Review

The overnight AI standup works in three layers:

Part A — Board-first agents

Configure your agent sessions to write board issues throughout the night as they complete tasks. Each completed task becomes a board issue titled with the work done, marked done. Issues that hit a blocker are created with BLOCKED: [reason] in the description. By the time you wake up, the board already reflects the overnight state without any special morning report — the standup summarizes what's already there.

Part B — Morning summary schedule

Create a scheduler entry for each session (or one master schedule that broadcasts to all sessions) that fires at 7am:

curl -sk -X POST -H 'Content-Type: application/json' \
  -d '{
    "title": "Morning standup — myproject",
    "session": "myproject",
    "command": "Summarize the last 8 hours of work. List: (1) completed tasks with file changes, (2) in-progress items, (3) any blockers or places you stopped and why, (4) what you plan to do next. Create a board issue titled \"Standup $(date +%Y-%m-%d)\" with this summary.",
    "schedule_expr": "daily at 7am"
  }' \
  https://localhost:8822/api/schedules

The agent receives this prompt, reflects on its session history, and posts a structured issue to the board before you check your phone.

Part C — Phone review

Open the amux iOS app or PWA. Tap the Board tab. You see the night's standup posts in order — one per session. Filter by today's date or search for "Standup" to find this morning's posts instantly.

What a Good AI Agent Standup Includes

Tell your agents exactly what to include in the standup. A complete standup post contains:

Completed tasks — what files changed and a one-line why
Tests passing / failing — current test suite status
PRs opened or updated — links to pull requests created during the session
BLOCKED items — explicit flag with the exact reason the agent stopped (missing credentials, ambiguous spec, conflicting requirement, merge conflict)
Next planned action — what the agent will do when you send it back to work

Add this to the end of every overnight agent prompt: "Post a standup summary to the board at the end of your session, or before your context runs out." That instruction ensures the report is posted even if the context window fills before morning.

The 5-Minute Morning Review Workflow

Open amux iOS app or PWA — the sessions view shows live status dots for each agent
Tap Board tab → scan for today's standup posts — search "Standup" or filter by date
Scan BLOCKED items first — these need your input before agents can continue
Move verified done items to 'verified' — mark confirmed completed work
Send steering instructions — open peek panel on any running session, use the steering bar or saved message chips to send next steps

Total time: 5 minutes. You never open a terminal, never attach to a tmux session, never sift through git log. The board is the single pane of truth.

What Happens When an Agent Is Blocked

A properly configured agent flags blockers explicitly — not as a missing commit, but as a board issue with "BLOCKED" in the title or description. Three blocker types to handle differently:

Ambiguous spec — you need to clarify the requirement. Send a steering message with the answer.
Missing credentials / access — you need to rotate an API key or grant access. Do it in your tooling, then tell the agent to continue.
Merge conflict — you need to merge the upstream changes. Resolve locally, push, then send "Pull the latest and continue."

The key is that the agent stops cleanly at a known point rather than silently hallucinating past a blocker or consuming tokens in a loop. Self-healing handles crashes and context overflow; blockers that require human judgment always surface to the board.

Standup Comparison

Approach	Time required	Blockers surfaced	Overnight visibility	Context depth
amux board morning review	5 min, async	✓ explicit BLOCKED flag	✓ full overnight	Full session output + board
Human daily standup	15–30 min, sync	✓	✗ no overnight	What people remember
GitHub commit digest	3 min	✗	Partial (commits only)	Code changes, no blockers
Slack bot daily digest	2 min, async	✗	Partial (message summary)	Surface-level only
Manual terminal check	15+ min	If you find them	✓ if you scroll enough	Raw terminal output

Frequently Asked Questions

How do I see what my AI agents did overnight?

Open the amux board on the iOS app or PWA in the morning. If standup schedules are configured, each session posts a summary to the board at the scheduled time (e.g. 7am). Each summary shows completed tasks, in-progress items, blockers, and next steps — all visible without opening a terminal.

Can AI agents replace daily standup meetings?

For an AI-only team, yes — each agent posts a structured standup to the shared board and you review it asynchronously. For a human+AI hybrid team, AI standup posts give the human team visibility into what their automated counterparts did, eliminating the need to check separate dashboards and reducing the cognitive overhead of tracking parallel work.

How do I schedule an AI agent to post a morning report?

Use the amux scheduler API or dashboard. Create a schedule with "schedule_expr": "daily at 7am" and a command that asks the session to summarize its work and post to the board. You can also configure it via the amux dashboard under the Scheduler tab — no terminal required.

What should an AI agent standup include?

Completed tasks (with files changed), tests passing/failing, PRs opened or updated, explicit BLOCKED items (where the agent stopped and why), and next planned action. Add "Post a standup summary to the board at the end of your session" to every overnight agent prompt — this ensures the report is posted before the context window ends.

How does the amux board work for team visibility?

The amux board is a shared SQLite-backed kanban visible to all connected clients — desktop browser, iOS app, and Android PWA. Agents create issues atomically via the REST API (SQLite compare-and-swap prevents duplicate work). Humans move issues from 'done' to 'verified' after confirmation. No separate project management tool needed for AI-only or human+AI teams.

Run your AI engineering team overnight — wake up to a full standup report

amux is an open-source control plane for running an AI engineering team from a single dashboard or your phone. Self-healing, single Python file, zero external dependencies, MIT licensed.

git clone https://github.com/mixpeek/amux && cd amux && ./install.sh
amux register myproject --dir ~/Dev/myproject --yolo
amux start myproject
amux serve  # → https://localhost:8822

⭐ Star on GitHub Get started →