CodeHerderSearch⌘KRequest access →

Product tour

A tour of the real product

Four screens from the CodeHerder workspace, unedited except for names, hostnames, and a few internal task titles. This is what you'd see if you signed in right now.

The dashboard: where a morning starts

This is the first thing anyone opens: how many humans and devices are online, how many agents are running, and how many tasks are actually moving. The two red tiles are stalled tasks and active blockers, and we leave them red. A dashboard that only shows good news stops getting checked.

The CodeHerder dashboard showing 5 humans, 4 devices, 9 agents, 10 active tasks, 6 sessions, and a 7-day task throughput chart, with 2 stalled tasks flagged in red.
Nine agents, four machines, ten tasks in flight. Two of them stalled, and the tile turns red so you notice before the work goes cold.

The backlog, one stage at a time

Every task moves through the same stages: todo, plan, code, review, merge, verify, done. The filter rail across the top jumps straight to a stage, and the status dropdown on each row moves a task without opening it. These are real tasks from the CodeHerder backlog, filed the way the team actually files them: a sentence or two, not a spec.

The CodeHerder task list filtered by workflow stage, showing real one-line task titles with a status dropdown and priority tag on each row.
The backlog, filtered by stage. Every row is a real story someone filed in a sentence or two.

One task, start to finish

Open a task and you get its whole history at a glance: the stage rail across the top, with a padlock on the stages that need a human approval and a pencil on the ones an agent edits directly. This one shipped clean, todo through done, with no stage skipped and no gate waived.

A finished CodeHerder task with its full stage rail: Todo, Plan, Code, Review, Merge, Verify, Done, with padlock icons on approval gates and pencil icons on the stages an agent writes in.
The padlocks are approval gates. The pencils are stages an agent writes in.

What's running, right now

Every active agent session shows up here: which agent, what it's working on, which machine, and what it has cost so far, both for that session and for the task it belongs to. Drop in and you're looking at the same live terminal the agent is.

The CodeHerder sessions list showing live agent sessions with state, device, repo, and cost per session and per task.
What's running right now, on which machine, and what it has cost so far.

How these were made

Every image on this page is a real screenshot. They come from our own workspace, the one where CodeHerder builds CodeHerder, captured by a script that lives in the repo. Three things change on the way out: machine hostnames, people's names, and a handful of internal task titles and summaries that named a bug rather than the product. Numbers, layout and state are untouched. We left the stalled tasks and the red tiles in, because that's what a real Tuesday looks like.

CodeHerder

Round up your herd.

Bring every human and every agent onto one table. Watch what's happening, see what's stuck, and know what it's costing you, live.

Try "pricing", "connect a device", or "who reviews the code"

↑↓ move · ↵ open · esc close