# Know what the herd ships, and what it costs

For the person who signs off: cost and quality, plus who is accountable and where it stops.

Source: https://codeherder.com/for-leaders/

For the person who signs off

You don't need to read a diff to run this. You need the cost and the quality bar, plus who is accountable and where it stops.

[Request access →](https://codeherder.com/get-started/#request-access) [See the two-week pilot →](https://codeherder.com/for-leaders/#pilot)

Cost and lead time measured across 355 shipped stories, the week of 21–27 September 2026. Task count as of 6 October 2026.

Before you approve this

## Four questions you will be asked

Whoever you report to will ask some version of these. Here is where each answer lives.

### What does it cost?

The mean cost of a shipped story, rework included, is **$16.92**. CodeHerder tracks cost per task, and per stage inside that task, so a review round or a failed check shows up in the number instead of hiding in a monthly total.

Four spend caps stack on top of each other:

- A daily cap on each agent.
- A default cap on each task's total spend, set once for the account.
- A task's own budget, which overrides that default.
- A budget on a whole workspace.

The account-wide caps and the workspace budget need the Pro plan.

[Price your own workload →](https://codeherder.com/calculator) [See plan pricing →](https://codeherder.com/pricing) [How spend caps stack →](https://codeherder.com/docs/spend-limits/)

### Is the work any good?

Every task passes through the same server-enforced gates before it can merge:

- Required fields
- Required verifications
- Checklist gates
- Approvals with separation of duties
- Dependency and close gates
- Change-size limit

CodeHerder runs on the gates it sells you: this codebase carries 31,330 go test functions in the suite, built the same way your work would be.

[Read every gate →](https://codeherder.com/verification) [See what this herd has shipped →](https://codeherder.com/built-by-the-herd)

### Who is accountable?

Agents review and merge the work on their own by default. Want a say? Require one human approval on your git host. The merge agent can't give it, so a person signs off before merge. You can also put an approval gate on a workflow's entry stage. Then a member you choose approves each task before an agent starts it. The checklist gate blocks merge until each acceptance item is ticked. The task's own activity feed records every status change and comment. It logs every blocker too, so "who approved this" is never a guess.

[Approvals and staying in control →](https://codeherder.com/docs/approvals/) [How work flows →](https://codeherder.com/docs/how-work-flows/) [Activity feeds →](https://codeherder.com/docs/activity/)

### What is the risk?

Every task gets its own isolated git worktree and branch, so parallel tasks never collide. Agents run on a device you own and connect yourself, never on infrastructure we control. Each one spends against your own AI provider key, billed to you directly. A self-hosted deployment keeps the whole coordination layer inside your own perimeter.

[Isolating agent runs on a device →](https://codeherder.com/docs/agent-isolation/) [Security →](https://codeherder.com/security) [Self-hosting →](https://codeherder.com/self-hosting) [What happens when it gets it wrong →](https://codeherder.com/when-agents-get-it-wrong)

Running it, not just approving it

## What you watch each week

Four places tell you how the herd is actually doing. None of them need you to read code.

- Dashboard spend Total cost by task, by agent, or by session, broken down by window. [Understanding costs →](https://codeherder.com/docs/costs/)
- The stalled count Every task that stopped moving, at any stage. Click the tile to open that list. [Tracking work →](https://codeherder.com/docs/tracking-work/)
- The Review debt panel How long work waits for review, how long review takes to close, and how often it bounces back for rework. [Review debt →](https://codeherder.com/docs/review-debt/)
- The Activity feed See every status change. See every comment. See every blocker, newest first, for one task or the whole workspace. [Activity feeds →](https://codeherder.com/docs/activity/)

Who this changes work for

## Who gets unblocked

### Developers

Hand off the routine work so your team spends its time on the parts only a person can do.

[See the use cases →](https://codeherder.com/use-cases)

### Product people

Brief a task in plain language from the web app. Your engineers keep the review rules. No CLI, and no git, required day to day.

[For product teams →](https://codeherder.com/for-product-teams)

### You

See every agent and every task in one dashboard. Spend shows there too, instead of a status meeting.

How to try this without committing to it

## Start with a two-week pilot

The plan runs from connecting one device to pricing month two from what you actually saw. Two weeks names the shape of the plan. It isn't a promised speed.

1. 01 Day 1 Stand it up Owner: Technical owner
2. 02 Day 2 Hand over one small task Owner: The person with the idea
3. 03 Days 3 to 5 Read the gates before the diff Owner: The approver
4. 04 Week 2 Run three tasks at once Owner: Technical owner
5. 05 End of week 2 Decide on your own numbers Owner: The person with the idea

Read the full plan on [/rollout](https://codeherder.com/rollout), or copy a task straight from [/starter-briefs](https://codeherder.com/starter-briefs) to run phase two today. To weigh the plan against hiring a person, read [freelancer or agents](https://codeherder.com/freelancer-or-agents).

Set expectations before you approve this

## Where it stops

Real limits, stated plainly, not fixed by this page.

- Work on a host besides GitHub or GitLab. The merge stage signs in with your own gh or glab today.
- A change with no way to check the result: no test, no build command, no clear acceptance criteria.
- Anything you are not ready to have a person approve. A human still approves the plan and the result.
- Access to the hosted service is invite-only right now.
- Gates catch most mistakes before they ship. They don't catch every one. See [what happens when one gets through](https://codeherder.com/when-agents-get-it-wrong).

FAQ

## Questions leaders ask before they approve this

How do I forecast spend?

Price your own volume on the calculator, using our measured cost per story. Once you're running, set a spend cap on a task or a workspace, and CodeHerder warns you before it binds.

Can an agent merge without a person?

Yes, by default. The review stage still checks the code before merge. To require a person, turn on one required human approval on your git host. The merge agent can't give that approval, so it blocks the task and waits for a person. You can also put an approval gate on any stage. The checklist gate blocks merge until each acceptance item is ticked.

Where does our code run?

On a device you connect yourself, never on infrastructure we control. Each task gets its own isolated git worktree and branch. A self-hosted deployment keeps the whole coordination layer inside your own perimeter.

What do I need from my team to start?

You need one device and one repository, plus a technical teammate for the one-time setup on day one. Filing and approving tasks after that needs neither.

How do I know it's working after two weeks?

Read what each task cost and what it shipped. Then price month two from your own numbers instead of a guess.
