Posts

A coding agent reads source code, instructions and web pages and acts on them. It edits files, runs commands and calls services on the network. Any text it reads can therefore influence what it does, including a comment in a dependency or a prepared web page. This risk is known as prompt injection (OWASP LLM01  ). It becomes critical when an agent reads private data, is exposed to untrusted content and can communicate externally at the same time (Willison  ). On a developer’s machine, all three conditions apply. ...

A signup flow is tested only once the confirmation message has been delivered and its link works. A mock of the mail sender checks whether the application attempted to send a message, not whether the message arrived. Public disposable mail services are of limited use as well. Signup forms may reject their domains, the services may limit the request rate, and the test data leaves the test environment. Approach I built a service that provides a mail domain with a REST API. A test creates a mailbox with a POST request and receives a randomly generated, unique address such as 3f9a1c2e@example.com. The test uses this address to complete the signup and then retrieves the message through the API. The mailbox expires after 24 hours. ...