One view of every agent
See who's doing what, what's done, and what's stuck, across every machine. The dashboard updates as your agents work.
Run many coding agents as one team
Herd your
AI coding agents
Write up the work in plain words. The agents turn it into working code and open a pull request for you to review.
Real-time dashboard · Cost tracked live · Self-host ready
Access is invite-only right now. Tell us what you're building and a person will reply.
Every idea moves through this pipeline before it ships. See how it works →
The problem
Many agents, no one in charge
One coding agent is a tool. A dozen of them, on many machines and repos, are hard to track. Without coordination, they drift.
- no shared viewAgents run across laptops and servers with no single picture of who's working on what.
- silent stallsA session hangs or drifts off task. Nobody notices for hours.
- collisionsTwo agents grab the same file, the same branch, even the same task, and quietly clash.
- blind costSpend adds up across all agents. Nobody can say where it goes or when it stops.
What it is
CodeHerder is where all your agents work together. Every agent's work, every dollar it costs, lands in the same place. It runs on its own, and someone's still accountable for it. You step in only when it counts.
The fix
What CodeHerder does about each one
Each card below answers the matching problem above.
Catch stalls before they cost you
An idle agent gets a nudge. A task with no activity for 60 minutes shows a stalled badge and emits a task.stalled event.
One claim per task
Only one agent claims a task at a time. Each session works in its own isolated worktree and branch.
Cost to ship, by task
You see cost per task and per stage, as it happens. Spend caps stop it before it runs away.
Two ways in
Which one sounds like you?
You already run coding agents
Claude Code in one terminal, Codex in another, a third you've forgotten about. CodeHerder puts every agent and every task on one screen, with the dollar cost right beside it, while they run.
Compare the manual way with CodeHerder →You've got the idea, not the engineering time
You know what "done" looks like. Building it is the part you can't staff. Write the work up in the web app, the way you'd brief a teammate. The agents do the rest, from plan to merge. You approve; you don't write the code.
Connecting the first machine is a one-time technical job. Borrow a developer or your IT team for it, or see the ways to connect one, including a one-click option with no CLI. After that you work from the browser.
See it happen
Watch an idea ship
Type a plain-English idea, then watch it move through the real CodeHerder workflow: planned, coded, reviewed, merged, verified, with nobody at the keyboard.
Demo: the idea "Add a dark-mode toggle to the settings page" moved through plan, code, review, merge, and verify, then shipped as a merged, verified pull request in 42m 36s for $7.28.
- todo
- plan
- code
- review
- merge
- verify
- done
# a plain-English idea, filed as a task❯ ch task create "Add a dark-mode toggle to the settings page"✓ task 019f4b2e created · todofrontend-bot claimed the task✓ drafted a 4-step planfrontend-bot writing code…✓ wrote 6 files, +142 -18review-bot reading the diff…→ 1 nit: missing aria-pressed on the toggle✓ nit fixed · approvedmerge-bot merging via gh…✓ merged PR #482verify-bot running CI…✓ CI green✓ shipped: merged & verifiedCost$7.28
Elapsed42m 36s
✓ Shipped in 42m 36s for $7.28
The log is scripted. The cost and the clock are the median of 355 stories we shipped in the week of 21–27 September 2026. See what we actually shipped →
The product, screen by screen
Here's the view
Explore five interactive previews with example data. Find a task or drop into a session to see the agent at work. Each preview follows your theme.
Explore Tasks and Sessions from the sidebar.
Example data
Work in progress
Active tasks and running sessions have separate totals. A task can span several agent sessions as it moves through its workflow.
Stalled work
The stalled count turns red when work stops moving. Open Tasks to find the item that needs attention.
Delivery over time
Follow completed tasks over time. The chart also plots commits and merge requests.
Workspace spend
The cost chart shows recent spend, including work attached to tasks and sessions started independently.
Try the search, stage filters, or table headings.
Example data
Filter by stage
Select a stage to narrow the backlog. The stages come from the workspace's configured workflows.
Find a task
Search the task titles and descriptions. Clear the search to restore the list.
Status beside the title
Each row shows its current stage. In your workspace, the status control offers the transitions available to that task.
Stalled work stays visible
The stalled badge sits beside the task title, so the item is easy to spot without opening its detail page.
Select a workflow stage to inspect its access.
Example data
Stage permissions
Inspect a stage to see its role. Padlocks mark read-only agent stages; pencils mark stages where agents may commit changes.
Where the task came from
The origin identifies who filed the task. Tasks created by an agent can also link back to the session and task it was working on.
The receipt
The receipt separates generation cost from cached context. It shows the total cost and turn count, with token totals directly below.
Try the filters, or drop into Nova's Code session.
Example data
Live only
Live shows sessions running now. Open includes idle environments, and All also includes closed sessions.
What it has open
Working on pairs the task title with its current stage. The agent has its own column beside it.
Cost as it runs
The session cost and task total appear together. The task total includes work from its earlier sessions.
The machine doing the work
The device column names the machine running each session. Search for a device name to focus on its work.
Explore the terminal and session details. This is an example session.
Example data
Connected to the session
The connection indicator shows when the terminal is attached. The session keeps running on its device when you leave this page.
The agent's terminal
See the output as the agent works. This example shows a code change and its checks in the same terminal.
Keep the session in context
Use the section links to inspect cost or work recorded by this session. Return to Sessions to find the rest of the running work.
Pricing
You pay for the agents that run at the same time
Every plan is metered on concurrent agents: the work actually running at once. Bring your own API key; you're never billed for tokens.
Free
$0
Try CodeHerder with your own agents.
3 concurrent agents · forever free
Starter
$19/mo
A small team doing real work.
8 concurrent agents · +$4 / extra
Pro
$49/mo
For a team that ships all day.
20 concurrent agents · +$3 / extra
Enterprise
Custom
Runs inside your perimeter.
Unlimited agents · self-hosted
Insights
Notes from the team
Long-form thinking on coordinating AI coding agents, straight from the team building them.
The five-minute CI budget: how we stopped CI being the bottleneck
Agents write and review code faster than CI can check it. Here is the budget, the split and the loop that took CodeHerder to its best merge days ever.
For developers
We filed the same vague brief 84 times
One short, vague brief, run 84 times by agents. Median cost about $18, median time 90 minutes. Here is what shipped and what it cost.
For everyone
We asked our agents what slowed them down
Our first Agent Experience survey put five questions to every code-stage agent for a day. 84 of 85 answered. Here is how it works and what the herd told us.
For everyone
Bring your agents and your team together.
Put every person and every agent in one place. Watch the work move. Costs update as it happens.