The Akita agent use eBPF to watch API traffic in blackbox way. Using eBPF makes it possible to drop into any system without needing to change code or use a proxy.
If you're curious about how we use packet filtering to watch traffic, you can read the following blog posts:
You can also check out our open-sourced CLI here.
After doing some basic analysis on the client side, the Akita agent ships request/response metadata off to the Akita Cloud for analysis.
In addition to tracking latency, throughput, and errors, the Akita cloud automatically infers:
- API endpoint structure. Akita uses structured, compiler-inspired techniques to automatically infers endpoint structure, including API path parameters.
- Request and response types. Akita users type inference on request and response fields to infer data types.
- Changes to API behavior. Akita uses previously inferred information, combined with API metrics, to report on changes to the API.
Our client only sends back request and response metadata to the Akita cloud. (In full detail: the Akita CLI uses a salted hash on payload data and never unhashes.) This means Akita never sees potentially sensitive data in your requests and responses. Read more in our Data Access & Handling Policy.
Updated 2 months ago