CodeArts Agent

The deciding trait is that it is one product wearing four faces. A standalone IDE, a command line with a text interface, and plugins for both editor families all sit on the same backend, so the engineer who wants a terminal and the engineer who refuses to leave their editor get the same behaviour and the same history. That consistency is worth more than any single feature.

The price of entry is an account and a relationship. Pick it if you already have both. Pick something you can install and forget if you do not.

Automatic delegation removes the one decision a user is actually qualified to make. When a system decides on its own how to divide a task, a bad division is invisible until the results arrive, and running several sessions at once means the same misjudgement is being paid for in parallel. Nothing published describes how a decomposition is reviewed before it runs.

What it does right is checkpoint. An undo exists for the result, which is the correct pairing for a loop the user did not plan, and it is documented alongside the delegation rather than buried.

1. The division is principled. A hosted index can afford to read a repository no laptop could hold, and a language server on the machine answers precisely the questions that must reflect the file as it is right now. Using one for breadth and the other for currency is the correct allocation of each mechanism's strength.

2. The stated scale is a capacity claim rather than a quality claim, and no retrieval accuracy is reported. 3. Feeding diagnostics into the edit path is the detail worth copying, since it makes the compiler an input rather than an afterthought.

A four-step consumer ladder from free to five hundred yuan is a company testing where the willingness to pay sits, and the metered layer underneath tells you what is really being sold. The parent runs its own model service and its own foundation models, so every hour spent in this editor is consumption on infrastructure it owns end to end. Vertical integration is the moat.

Likely acquirer: none, and no exit is contemplated. Position: the longevity risk here is close to zero and the concentration risk is the mirror image of that, since one company owns the client, the model and the cloud.

The controls exist and they are the right controls. Assets are governed at enterprise, team and individual scope, so a rule my architects write is not something sixty engineers can quietly override, and quotas cap the meter before finance discovers it. An address allow-list and repository isolation answer two questions my security team always asks.

The gap is the contract, since amounts are published in one currency on one portal and my procurement runs in another. Approved with conditions: a quote in my currency, a written retention policy, and directory provisioning confirmed before rollout.

Being a server is the part I respect. Its static analysis is exposed as a tool other agents can call, which means one piece of this stack is reusable outside the product, and that is more generosity than a vendor of this size owes anyone. My own servers attach as clients too.

Everything else runs on somebody else's terms. Nothing serves weights from my hardware, the source is shut, and the moment my employer pays for the higher tier, my model picker becomes an administrator's setting. I can configure this thing. I cannot own it.