
Capy's software engineering platform coordinates coding, testing, and review agents across cloud and local environments.
Capy's workflow combines implementation with CI feedback, pull-request review, and recurring automations. Parallel subagents can divide work across independent or shared virtual machines, giving teams a coordinated way to run several development tasks.
Customers can select supported models and connect their own subscriptions, API keys, or gateways. The offering also includes unlimited members and integrates with GitHub, Slack, and Linear, allowing work to begin and receive updates through existing team tools.
Capy's official changelog dates its first release to August 11, 2026. The initial offering ran coding threads on cloud machines with cloned repositories, opened pull requests, and responded to CI completion, reviews, and merge events.
The macOS application first shipped on September 7, followed by Windows and Linux support in the September 10 release. Later September updates expanded model connections, pull-request reviews, and MCP settings within the same Capy offering.
Capy offers Lite at $20 monthly, Pro at $100, and Max from $200 to $1,000, alongside custom Enterprise plans. Annual billing reduces the stated monthly equivalent by 20%, and the introductory Lite offer costs $1 for seven days.
Plans include unlimited members and credits shared across the organization, with higher tiers increasing cloud-machine capacity. Credits pay for model usage and machine time; customers can add balance or connect supported subscriptions, API keys, and gateways.
Capy is a coding-agent workspace for implementing software changes and producing pull requests. A thread starts from its web interface, Slack, Linear, or API, then works with cloned repositories and responds to CI results and review feedback.
The offering combines parallel subagents, pull-request reviews, scheduled automations, and configurable development environments. Its workspace connects repositories and collaboration tools, while desktop applications extend threads to local computers.