Slack Plan and Task Card Blocks: Every Status, Update and Limit Tested

The task_card block shows one step an app or agent is working on, with a status icon, optional details and output, and source links. The plan block groups up to 50 such tasks under one title. Slack added both in February 2026 and changed how they look on 20 August 2026. On 3 October 2026 we posted them from a test bot in our own workspace, in Slack web, and tried every status, an update, and bad values. A standalone task_card rejects pending, while a plan accepts it.

The blocks we posted

import json
from lab import raw   # our helper: POSTs JSON to https://slack.com/api/<method> with the bot token

DM = "D0C6CRJSZK8"
def rt(t): return {"type": "rich_text", "elements": [{"type": "rich_text_section", "elements": [{"type": "text", "text": t}]}]}
def task(i, status, title=None, **kw):
    d = {"task_id": f"t{i}", "title": title or f"Task {i}: {status}", "status": status}
    d.update(kw); return d

tasks = [task(1, "pending", "Read the 4.53 changelog"),
         task(2, "in_progress", "Group changes by area", details=rt("Sorting 38 entries")),
         task(3, "complete", "Find breaking changes", output=rt("2 breaking changes found"),
              sources=[{"type": "url", "text": "docs.slack.dev", "url": "https://docs.slack.dev/changelog"}]),
         task(4, "error", "Fetch the GitHub release", output=rt("HTTP 404 from the release URL"))]
r = raw("chat.postMessage", {"channel": DM, "text": "617 plan test",
        "blocks": [{"type": "plan", "title": "Release summary plan", "tasks": tasks}]})

cards = [{"type": "task_card", **task(i + 10, s)} for i, s in enumerate(["pending", "in_progress", "complete", "error"])]
r2 = raw("chat.postMessage", {"channel": DM, "text": "617 cards test", "blocks": cards})
cards3 = [c for c in cards if c["status"] != "pending"]
r3 = raw("chat.postMessage", {"channel": DM, "text": "617 cards test 3", "blocks": cards3})

The task card reference lists only in_progress, complete and error. The plan reference's own example uses pending, and Slack accepted it there.

How each status looks in Slack

Slack showed the plan collapsed to one line, "Release summary plan", with a spinner because a task was still running. Clicking the title opened it:

What each status got:

Inside a plan, pending and complete look the same. The plan's header icon is what tells the reader the plan is done. details showed under the title, output under that, and sources as links. Tasks with details or output got an expand arrow.

Updating the plan with chat.update

We sent the same plan again with every task set to complete and a new output on the failed one:

for t in tasks:
    t["status"] = "complete"
tasks[3]["output"] = rt("Found it on the second try")
raw("chat.update", {"channel": DM, "ts": plan_ts, "text": "617 plan test",
     "blocks": [{"type": "plan", "title": "Release summary plan", "tasks": tasks}]})

The call returned {"ok": true}. The plan collapsed again, and its header spinner became a check mark. Opened, it read:

So to show progress, send the full plan each time. There is no call that changes one task; chat.update replaces the whole block. chat.startStream also accepts task_update chunks, which we did not test; our chat.startStream test covers the text side.

Limits and errors

Each row is one chat.postMessage with a bot token:

We found no title limit. Slack accepted 200,000 characters and showed the long titles in full, wrapped across the screen, so cap titles in your own code. details and output must be a single rich_text object, the same structure as the rich_text block.

We did not check the mobile app. When a block fails, our invalid_blocks guide explains how to read the json-pointer.