Cursor has given its cloud agents a place to work. The AI coding company rolled out development environments on 13 May that let a single configuration hold all the repositories an agent needs, reusable across sessions. That matters more than it sounds, because an agent is only as useful as the environment it runs in.
Disclosure: RECATOOLS is written with the assistance of Claude, made by Anthropic, which is also named in this article. Readers should weigh that interest.
What shipped
Cloud agents now support multi-repo environments, so one setup can span the several codebases a real task touches. Docker builds got a layer-caching upgrade: change the Dockerfile and only the altered layers rebuild, which the company says makes cache-hit builds 70 per cent faster. And build secrets let an agent reach private package registries during a build without those secrets leaking into the running environment.
The quality-of-life bits
Cursor will now interview you as it sets up an environment, asking questions, flagging missing credentials and checking the configuration works before an agent starts. Each environment keeps its own version history, so a team can review changes and roll back. For teams that would rather not write a Dockerfile, Cursor can inspect the repos, work out the dependencies and generate one to edit.
Why it matters
What separates a coding agent that demos well from one a team trusts on a live codebase is mostly setup. The agent needs the right repos and the right secrets, and a build that does not take ten minutes. A new model draws more attention, but closing that gap does more to decide whether agents get used. Cursor is competing with GitHub, Replit and a field of cloud players on this ground, and the winner will be whoever makes the boring parts disappear.
Three months on, the plumbing turned out to be the asset
This article argued in May that environment setup matters more than model quality in deciding whether a coding agent gets used on real work. What has happened to Cursor since suggests the industry priced that plumbing considerably higher than anyone was saying at the time.
On 8 July, SpaceXAI released Grok 4.5 at US$2 per million input tokens, a coding-focused flagship trained with Cursor data. Six weeks after that, SpaceX closed its acquisition of Anysphere, Cursor's maker, for US$60bn paid entirely in SpaceX stock. It is the largest startup acquisition on record, and Cursor now sits inside a SpaceXAI division.
The development environments did not cause the acquisition directly. What they did was let cloud agents run more real tasks against more real repositories, and the record of those runs is the kind of data a coding model is trained on. Cursor's maker spent its engineering effort making the boring parts work, and a rocket company paid US$60bn in stock for what that effort produced.
The competitive ground moved
The contest named in May was Cursor against GitHub, Replit and a field of cloud players, with the win going to whoever made the boring parts disappear. Those boring parts are now being commoditised, so the contest is being fought on other ground.
Warp shipped Factories on 20 August, running coding agents as a pipeline rather than a prompt: triage, spec, implementation, review and verification as separate stages, with a different model permitted at each one. Model neutrality is the explicit selling point, and Warp made it days after the Cursor deal put Cursor's own neutrality in doubt. For Cursor, the question has become whether an editor owned by a model company will keep routing to rival frontier models.
Cognition raised US$1bn at a US$26bn valuation with Devin on a US$492 million annual run-rate. Every player named in May is still standing, but the contest is increasingly about who controls the model underneath.
The bottleneck moved to review
Solving setup has exposed the next constraint.
Warp's own framing is that a staged pipeline relocates the bottleneck to review capacity. The evidence on how well that review works is not encouraging. A permission-approval study logged 409,000 decisions across more than 40,000 runs and found reviewers missed 33.7 per cent of malicious commands while blocking safe ones at rates as high as 59 per cent. Anthropic's own figures put the human catch rate at 13.6 per cent against a classifier's 89 per cent, and from 14 August Claude Code ran new Pro, Max and Team sessions in auto mode by default, with a classifier rather than a human approving most actions.
The gap between a demo-ready agent and one trusted on a live codebase was mostly about setup in May. Setup was largely solved by August, and the trust question had moved to a human review step that measures badly in both directions, prompting at least one vendor to replace the per-command prompt with a classifier.
What a reusable cloud environment costs
A consequence not visible in May is that an environment living in the cloud, reused across sessions, concentrates a team's dependency on its host and on the registries it pulls from during a build. GitHub was degraded for about three hours on 17 August, with archive downloads failing at roughly half, and continuous integration broke in organisations that had never registered they depended on GitHub being available. An agent whose environment rebuilds from a hosted configuration inherits that exposure by design.
Reusable environments are still worth having. The cost is that the reliability question has followed setup into the cloud, so a failure that once stopped one developer's local machine can now stop everyone who shares the dependency.
Correction, 18 October 2026: An earlier version said GitHub's Copilot Workspace reached general availability at Build 2026 and that Microsoft would move Copilot to a model called Project Polaris. Neither claim could be supported and both have been removed. The description of Claude Code's August default change has also been made more precise.
View author profile → · Editorial policy
