Much of modern programming involves using APIs. But APIs also cause developer pain: any code change in one service could break other services in nonobvious ways. And the changes could be as small as a type or data format change: something a developer would have caught easily in a simple monolithic application.

Akita catches breaking changes by watching API traffic to give developers structured behavioral diffs. Akita sniffs network traffic to generate API models, captured as annotated API specs. The API models allow Akita to alert on breaking changes precisely and accurately. Integrate Akita with source control/CI, canary deployments, and/or production with no code changes or proxies to get a semantic diff of API behavior.

You can install and integrate Akita in just minutes. Sign up to join our private beta today!

Watch a video of our demo at the NY Enterprise Tech Meetup here. (We were the opening act for Okta CEO Todd McKinnon!)

Akita helps you move faster

• See where you've drifted from the plan. Akita models API behavior as annotated specs, allowing you to diff across them or against existing specs to detect schema evolution.
• Understand actual usage of your API. By running Akita in production, you can see how the "effective spec" of your APIs differ from the intended specs.
• Catch breaking changes before they reach your customers. Get alerts about regressions in test/CI, canary deployments, or production.

By diffing on API models rather than code, Akita is able to tell you:

• Endpoints added that expose new functionality
• Parameters removed that may break existing clients
• Simple type changes (ex string to int) that can pollute data pipelines
• Complex type changes (ex phone number to datetime) that can break dependencies
• 👷🏻‍♀️ Coming soon: Implicit API Contracts

Akita works by watching API traffic

Akita sniffs network traffic and generates behavioral models of your API, including the endpoints, types, and specific data formats. To catch breaking changes, Akita diffs on these models instead of diffing syntactically on requests/responses, or—heaven forbid—source code. This allows Akita to pinpoint potential issues much more precisely.

Akita works through BPF filters, rather than an SDK or proxy, making it easy to run Akita in test or production environments—and even compare behaviors across the two. Here's a diagram of how the server-side version of Akita works.

Using Akita

Try out Akita directly from the command line, or integrate with your CI and/or deployments.

🔎 Collect API Traffic

Either use the Akita agent's apidump command to collect API traces, or generate HAR files from a standard browser or proxy tool (for instance, Charles).

🏗 Build API Models from API Traces

Use the apispec command to send API traffic traces to the Akita cloud for building API models. Akita stores API models as annotated API specs, showing you endpoints, fields, data types, and more. Traces contain only request/response metadata and never contain any property or parameter values, ensuring that Akita never sees your sensitive data.

🔮 Use API Models for Diffing and More

You can now use Akita API models for regressions, understanding schema evolution, or documentations.