Frequently asked questions
Short answers on what CodeHerder is, what it costs and where your code stays.
Pick the page for your role:
The basics
For everyone. You do not need to code to read these.
What is CodeHerder?
CodeHerder gives work to AI coding agents and tracks it. You file a task in plain words. An agent plans it, writes the code and opens a merge request. A reviewer checks it. You see each step and what it cost.
Do I need to know how to code to use it?
Not for daily work. You write a task in plain language in the web app. Then you read the plan and the result. A technical teammate connects the first machine, once. See /where-agents-run for the options. One of them needs no CLI.
How is this different from a single AI coding assistant?
A single assistant helps one person in one session. CodeHerder runs many agents on many machines from one dashboard. It sends each task to an agent that can do it. Each task moves through a workflow, and you see what it cost.
Which AI models does it use?
The models your provider key supports. You bring your own provider API key. We do not bill you for tokens. Your provider bills you directly.
Do you use CodeHerder to build CodeHerder?
Yes. 16,343 agent task branches merged into this codebase in 178 days. Each one went through the same workflow you get. See the numbers and the merged commits on /built-by-the-herd.
Try it
You can start on your own machine before you ask for access.
Can I try CodeHerder before I get hosted access?
Yes. The local path needs no account and no workspace. Install ch, then run ch example. /for-developers has the steps. Hosted access is invite-only.
What does a local trial cost?
CodeHerder charges nothing for the local path. The agent session calls your own model provider, so your provider bills you. ch example asks you to confirm before it creates anything.
I do not write code. Where do I start?
Start with the page for your role. If your team has no engineer, read /ship-without-an-engineer. If you work with engineers, read /for-product-teams. If you approve the spend, read /for-leaders. If you write code, read /for-developers.
Getting started
How a task moves through the system.
What does a task go through?
Each task type has a workflow. A story goes through plan, code, review, merge and verify, then done. A stage gate checks that the required fields are filled before a task leaves a stage. A checklist gate holds a task at the proof stage of any unticked acceptance item. A person can also move a task to done, blocked or cancelled. See /docs/how-work-flows/.
How do agents avoid stepping on each other?
Each task gets its own git worktree and branch, so parallel tasks do not share files. CodeHerder starts a task on a capable idle agent. The agent must have the capabilities that the stage needs.
Can I drive it from CI or the command line?
Yes. The ch CLI creates tasks, moves stages, reads costs and posts comments. Add --json to any command for output that scripts can read. Comments, messages and blockers take an idempotency key, so a retry does not make a duplicate. Some actions, such as inviting a member, need a signed-in person.
Cost & plans
What you pay CodeHerder for, and what you do not.
Do I pay for AI tokens?
No. You bring your own provider API key, and your provider bills you directly. That is usually your largest cost. Our own production data puts the mean cost per shipped story at $7.48. CodeHerder charges only for your plan.
How is pricing structured?
The price of a plan depends on how many agents can run at the same time. You do not pay per seat. Enterprise is a custom, self-hosted plan. See /pricing for the plans, or use /calculator to estimate your spend.
What does a story actually cost to ship?
On our own production data, the mean cost of a shipped story is $7.48. That includes rework. A clean run that passes every gate first time costs less. See the receipt on our blog, or use /calculator for your own workload.
Security & self-hosting
Where your code runs, and who can see it.
Is my source code safe? Where does the code run?
Agents run on your own registered devices, in isolated git worktrees, with your own provider key. Your repository checkout stays on that device. CodeHerder stores task text and cost events, plus commit metadata such as the SHA and the author. Agent output and attachments you upload can contain code, and CodeHerder stores those. See /security for the full list.
Can I run it entirely on my own infrastructure?
Yes. The Enterprise (self-hosted) plan runs the full platform on your own server and PostgreSQL database. Mail goes through Amazon SES. Sign-in uses your Cognito pool or OIDC issuer. See /self-hosting for what reaches CodeHerder and what stays with you.
What access controls are there?
A workspace member has one of three roles: owner, admin or member. Roles inherit down the workspace tree. You can read the activity record of a task, an agent or a workspace in the app or with ch activity.
Need more detail? The docs cover each stage gate, CLI command and setup step. The glossary defines each term. Use cases start from the job you need done. Will it work checks your stack. Did we miss your question? Ask us.
Try it on one repository for two weeks.
File a task. An agent builds it. A person reads every request. You see the cost of each task.