Board — Dashboard Tabs

The Board tab is an agentic kanban board. Work moves through columns, and each column can execute Heym workflows. A card is not just a task — it is a persistent agentic job that carries context, conversation history, execution state, outputs, and workflow runs.

Boards and columns

You can create as many boards as you need. A new board starts with the default columns Backlog → Planning → Development → Done.
Every board needs an Agentic Kanban Model (a credential plus a model). It is mandatory — the New board dialog will not create a board without one, and a board that has none keeps its actions disabled — because it is the model that maps each card into the inputs of the workflows it runs and turns their output back into readable text.
The gear icon in the board header opens the board settings: name, description (shown under the header) and the Agentic Kanban Model.
Columns are fully editable: add, rename, recolor, reorder, or delete them from the column settings (a column must be empty before it can be deleted).
Cards are ordered vertically by priority — drag to reorder.
You can also create a card from the Chat tab with a natural-language request. Chat asks you to select a board when needed and always adds the card to that board's first column.

Board settings also share the board with users (by email) and with teams, with read or write permission:

read — the board and its cards are visible, but nothing can be changed: no new cards, no moves, no comments, no runs.
write — full use of the board: cards, moves, comments and runs.
Only the owner sees the gear and delete buttons, and only the owner can change the board's settings, its Agentic Kanban Model or its shares.

Chains on a shared board always run with the owner's credentials and Agentic Kanban Model, so collaborators never need their own.

Card attachments

Open a card and use Attach file to add files to it — drop them on the box or pick them with the button, and remove them from the same list. Attachments are stored in Drive.

Every run resolves the attachments before the workflows start, by type:

Documents (pdf, markdown, csv, json, text) are extracted to text, capped at 20,000 characters per file, in the attachment's text field.
Images are handed over as a url the vision path loads directly (an LLM node's image input accepts it as-is).
Anything else is passed through as a plain reference (name, url, mime_type).

Workflows read them at $input.card.attachments (and always at $input.board.attachments, whatever the mapper mapped). The Agentic Kanban Model sees the extracted text and the image URLs too, so it can map an attachment straight into the field a workflow expects.

Column workflow chains

Each column can be configured with an ordered list of workflows (the chain). When a card moves into a column that has a chain:

The chain starts automatically in the background — the move itself never waits.
Workflows run sequentially. Each workflow's output is appended to the card context before the next one starts.
If a workflow fails, the chain stops: the remaining links are skipped and the card turns red.
When every link succeeds, the card turns green. While the chain is running the card pulses amber; a run paused for human review shows a static amber state.

Runs are recorded in Execution History with the board trigger source, and each card keeps its own run list with per-workflow outputs and errors. Open a card while a workflow is running and choose Open live beside the run to attach the editor to that exact execution. The same canvas animation, incremental Debug logs, and final result appear without starting a second run.

What workflows receive

The chain passes a standard payload as the workflow's input. Read it with normal expressions such as $input.card.title:

{
  "triggered_by": "board",
  "rerun": false,
  "card": {
    "id": "…",
    "title": "Write launch email",
    "content": "Draft the launch email for the beta",
    "metadata": { "attachments": [{ "name": "brief", "url": "…" }] },
    "comments": [{ "author": "user", "content": "…", "created_at": "…" }],
    "history": [{ "kind": "event", "content": "…", "created_at": "…" }],
    "previous_outputs": [{ "workflow_name": "…", "output": {}, "finished_at": "…" }]
  },
  "board": { "id": "…", "name": "…" },
  "move": { "from_column": "Backlog", "to_column": "Planning" },
  "chain": { "position": 0, "length": 2, "previous_workflow_outputs": [] }
}

comments — every comment on the card, oldest first.
history — the full activity timeline (comments, moves, outputs), capped at the most recent 200 entries.
previous_outputs — outputs from earlier completed runs (previous columns and follow-up rounds).
chain.previous_workflow_outputs — outputs of earlier links in the current chain.
On follow-up rounds move is null and rerun is true.

The planning loop (follow-up rounds)

A common pattern is to put a planning workflow on the second column (by default named Planning — the name does not matter; the slot does). That workflow enriches the card and asks follow-up questions:

Move a card into the second column — the workflow writes an enriched plan and its questions back to the card as an output.
Answer in the card's comment thread. The answer releases the card: it moves on to the next column, which picks it up with the answer in context. The second column's chain is not run again.
To improve the plan in place instead, press Run follow-up round. The same chain runs again with all accumulated context, then the card waits in that column until you add a comment to release it.

The planning gate is positional, not name-based: the leftmost column and the column immediately to its right never move on by themselves — they run their chain and wait there for you (a comment releases the second column). From the third column on, a successful chain advances the card to the right on its own, all the way to the last column. Renaming or removing the default "Planning" label does not change which slot is the gate. A card can only have one active run at a time.

Human-in-the-loop and Codex questions

If a chain workflow pauses — on a Human-in-the-Loop node, on an agent's HITL tool, or because a Codex node needs more information — the run is persisted as pending and the card shows the amber pending state. Each pause keeps its own answer surface: a HITL review link for HITL, the Codex follow-up screen for Codex.

Once you answer, the chain resumes on its own: the paused workflow finishes, its output lands on the card, the rest of that column's chain runs, and the card advances as usual. If the resumed run fails, the card turns red and the remaining links are skipped.