CodeHerder
Command your fleet of coding agents.
What problem does it solve?
Running many AI coding agents at once creates blind spots. You can't see what each agent is doing. Agents collide on the same files, and cost sprawls across machines and repos.
Who is it for?
Engineering teams and studios operating fleets of AI coding agents across multiple machines and repositories.
Why did we build it?
We run AI coding agents on every client engagement, often several at once across different repos. Early on, that meant a wall of terminal tabs and no shared view of what any agent was doing. We lost more than one afternoon untangling two agents that had edited the same file at once.
The cost problem showed up just as fast. Agent time isn't free. Once a fleet runs across machines, the spend scatters exactly like the visibility does. There is no single place to see what a task cost until the invoice lands.
CodeHerder is the tool we built to fix our own operation first. It gives us one place to see every agent, on every machine and repo. Tasking can't collide, and it tracks cost per task instead of reconstructing it after the fact.
How does it work?
- A shared task registry, not shared state
Agents don't need to share a filesystem or a process to avoid stepping on each other. They claim tasks against a central registry instead. That way, two agents can never be handed the same file at the same time, even on different machines.
- Collision detection at the task level
Before an agent starts, CodeHerder checks what else is in flight against that repo and file set. A conflicting claim gets rejected up front instead of surfacing later as a merge conflict or clobbered edit.
- Per-task cost ledger
Every agent run reports its own spend back to the task it was assigned. Cost then rolls up by task, repo, or machine, instead of arriving as one undifferentiated provider bill.
- One dashboard across every repo and machine
The registry is the same one every agent reports to, regardless of which machine kicked it off. That makes the fleet view a query, not a manual roll-up.
What's the hard part?
Collision-free tasking without shared state was the genuinely hard part. Agents run as independent processes, often on different machines, with no lock file or shared memory to coordinate through. So a central claim-and-check step has to enforce "don't let two agents touch the same file." That step has to be fast enough not to slow an agent down. It also has to be consistent enough that a claim made on one machine is instantly visible to an agent starting on another.
What does it do?
- Real-time visibility into every agent across machines and repos
- Collision-free tasking so agents never clobber each other's work
- Per-task cost tracking from a single dashboard
- Coordinate agents across many repositories at once
What does this prove we can do for you?
We build the same discipline for clients running their own agent fleets or multi-step workflows in production. It combines a central coordination layer, collision detection, and per-unit cost accounting.
Where this shows up in client work
Questions, answered.#
The questions we get asked most, answered plainly: no hedging, no marketing copy.
01 What is CodeHerder for?
CodeHerder coordinates multiple AI coding agents running across different machines and repositories at once. You see what every agent is doing, and agents never collide on the same files. You track what each task costs from one dashboard.
Link to this answer: What is CodeHerder for?02 Does it work with any AI coding agent?
CodeHerder coordinates at the task and repo level, not by controlling an agent's internals. It's built to sit alongside whichever coding agents you already run.
Link to this answer: Does it work with any AI coding agent?03 Do I need to change how my agents work to use it?
No. You don't rewrite your agents; they claim tasks against CodeHerder's registry. The coordination happens around them, not inside them.
Link to this answer: Do I need to change how my agents work to use it?A coordination platform for running many AI coding agents across machines and repos. It gives real-time visibility, collision-free tasking, and per-task cost tracking from one dashboard.
This is the same discipline, real architecture decisions and an honest account of the hard part, that we bring to client engagements.