CodeHerderSearch⌘KRequest access →

Rollout plan

Your first two weeks with the herd

This plan takes about two weeks. Go faster or slower as you like. It does not promise how fast a task ships.

Who does what

Three roles, split however your team likes

One person can hold more than one of these. Make sure someone covers each one before Day 1. If you're the one signing off on this rollout, read the case for it before you read the phases below. A product manager with a team can start at for product teams.

The person with the idea

Writes what "done" looks like, and files the first tasks. Answers a question when an agent asks one.

Doesn't need to read code, or touch the CLI.

When an agent needs your input →

The technical owner

Connects the first device and the first repository. One-time setup.

Doesn't need to stay involved once that first setup is done.

How do I add a device? →

The approver

Signs off where a gate asks for it. Answers agent questions on the tasks they approve.

Does not need to read every diff. An agent review stage checks each change before merge and comments on the task. Read that comment first.

Approvals & staying in control →

The five phases

Day 1 through the end of week 2

Read each phase before you start it. What actually happened decides what comes next.

  1. Day 1 · Technical owner

    Stand it up

    Connect one device and one repository, so an agent has somewhere to work.

    Sign-up is invite-only right now. Request access before you start.

    • Sign in and open Set up → Devices.
    • Set up agent isolation, then run ch device-server on a machine that already builds and tests your code.
    • Add one repository under Set up → Repos.

    Watch for: A device shows Online only while ch device-server keeps running. A machine that sleeps goes Offline within about a minute.

    Done when:

    • The device shows Online.
    • The repository appears under Set up → Repos.
  2. Day 2 · The person with the idea

    Hand over one small task

    File one bounded, checkable brief. Watch the herd build it.

    • Pick one small, well-understood job from the starter briefs.
    • Copy the brief. Edit only the parts specific to your repo.
    • File it as a task, then watch it move through plan and into code.

    Watch for: With a vague brief, the agent asks what "done" means. The task then moves to Blocked with a question. Answer it on the task page.

    Done when:

    • The task reaches review with a pull request or merge request attached.
    • The acceptance criteria are checkable against the real result.
  3. Days 3 to 5 · The approver

    Read the gates before the diff

    Learn what each stage's gate checks for you, before you widen the work.

    • Read the hand-off comment review leaves on the task.
    • Check what an approval gate holds, and what it does not.
    • Read what verify confirms before a task reaches done.

    Watch for: No built-in workflow has an approval gate. Add one to the entry stage (Starter plan or higher). Or turn on a required approval on your git host.

    Done when:

    • You can name what each gate checks before it lets a task through.
    • You know which stage, if any, needs your sign-off before it starts.
  4. Week 2 · Technical owner

    Run three tasks at once

    Route more than one task to more than one agent, without them colliding.

    • Give a second agent a capability label (Starter plan or higher).
    • File three tasks that touch different parts of the repo.
    • Watch each one work in its own isolated git worktree.

    Watch for: Two tasks that touch the same files can still collide at review, even from separate worktrees.

    Done when:

    • Three tasks run at once, each claimed by a distinct agent.
    • Each task's change stays inside the part of the repo it was meant for.
  5. End of week 2 · The person with the idea

    Decide on your own numbers

    Read what the first two weeks cost and shipped. Then price the next month from your own data.

    • Read the per-task cost on each task you ran.
    • Estimate next month's spend with the calculator, using your own rate.
    • Pick a plan. On Pro, also set a task budget or a spend cap.

    Watch for: After one send-back, the next code run asks for an Opus-tier agent. Rework costs more for that reason.

    Done when:

    • You can state what the first two weeks cost, in your own numbers.
    • You have a plan for month two, priced from what you actually saw.

What to hand over first

Start with a job that's easy to check

What to keep back, for now

Wait before you hand over these jobs

  • Work on a host besides GitHub or GitLab. Only github.com and gitlab.com are known hosts today.
  • A change with no way to check the result: no test, no build command, no clear acceptance criteria.
  • Anything that needs a person to approve it. No stage waits for a person by default. Turn on a required approval in branch protection. You can also add an approval gate to the entry stage (Starter plan or higher). It holds work before an agent starts.

See /will-it-work and /when-agents-get-it-wrong for the full list of limits.

What to measure

Four things CodeHerder already reports

CodeHerder sets no target number and no benchmark. Read your own two weeks. Then decide what good looks like for your team.

Per-task cost

ch task show <taskId> prints what a task has cost so far. It also shows where the task stands against any budget you set.

Understanding costs →

Change shape

Every task that changes a repository reports its change shape: how many files, lines, and directories it touched.

Change shape →

Stage history and rework rounds

Each stage transition is on record, including a send-back into code and which stage sent it there.

How work flows →

Blocked reasons

A blocked task carries a reason: who filed it, and when. Some clear on their own; others need you to act.

Blockers and blocked tasks →

FAQ

Questions people ask before Day 1

Do I need to follow this in exactly two weeks?

No. This plan takes about two weeks. Go faster or slower as you like. Nothing here times you.

Who has to approve the first task?

Nobody, unless you set an approval gate yourself. No built-in workflow has one. A gate on the entry stage lets a person approve before an agent starts. A required approval on your git host covers the merge.

What happens if the first task goes wrong?

Review usually catches it before merge. After a send-back into code, the next run asks for an Opus-tier agent. A task gets 3 send-backs by default. Then it moves to Blocked with a note, and a person decides what happens next.

Can I run this rollout without an engineer?

Day 1 needs a technical teammate for the one-time device and repo setup. Filing and approving tasks after that needs neither.

CodeHerder

Ready to start the two weeks?

Tell us what you're building, and we'll get your first device and repo connected.

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

↑↓ move · ↵ open · esc close