synlace / ferret

What is ferret?

Ferret is a web-based HTTP proxy that intercepts traffic, stores it in a local SQLite database, and pairs it with an AI assistant. You route your browser or app through Ferret's proxy port, and every request/response gets captured with full headers, body, and timing data. The TypeScript UI lets you browse history, search requests, and annotate findings. The Python backend integrates with multiple AI providers so you can ask questions about your traffic in natural language, and the system will generate pytest scripts that replay specific requests against your API.

Why is it gaining traction?

The key differentiator is zero-infrastructure storage. Most HTTP proxy web UIs expect you to run Elasticsearch or a similar heavyweight backend. Ferret uses SQLite instead, which means you spin it up with docker-compose and it just works. No JVM, no Kafka, no Ops headache.

The AI integration is the second hook. When you spot something interesting in captured traffic, you can open a chat session scoped to a specific request, host, or your entire history. The AI can search your request archive, write tests, and run them inside an isolated sandbox container. This turns passive traffic inspection into active API testing.

The Snare feature adds programmable request interception. You configure rules by method, host, path, headers, or body patterns, and matching requests are held for review before being forwarded or dropped. This is useful for security testing workflows where you want human approval on sensitive operations.

Who should use this?

Backend developers debugging APIs will find the traffic replay and annotation features practical for tracking down issues. Security testers working with AI-assisted tooling will appreciate the rule-based interception and automated test generation. QA engineers writing integration tests against staging environments can use Ferret to capture real request shapes and convert them into repeatable pytest suites.

Verdict

Ferret is a well-structured concept at a very early maturity stage (22 stars). The test coverage is thorough, the architecture is clean, and the SQLite-first approach removes real friction that alternatives impose. That said, production users should monitor the project for documentation and community growth before committing to it as a primary tool. The 0.850 credibility score reflects solid engineering practices but limited adoption -- worth evaluating for medium-complexity API testing workflows, but treat it as a promising early-stage project rather than a mature solution.