Writing a custom Rust URL router is squarely in AI's coding wheelhouse.

The task is structurally deterministic: parse a pattern, compile a regex, match a path, extract and validate named groups. The same logic applies every time, with no instance-by-instance judgment required.

Success criteria are concrete and testable: routes match correctly, parameters are extracted and typed, 404s and binding errors return clear messages. A non-human can verify correctness with unit tests.

The agent needs only Rust's standard library and the regex crate (or pure std if avoiding all third-party deps), both of which are well-documented in training data. No external APIs, credentials, or live context are required.

Bugs in a router can cause incorrect routing or security-adjacent issues like parameter confusion, but the output is code that must be reviewed and tested before deployment, making errors reversible and catchable before they cause real damage.

Design choices like error message wording, parameter type extensibility, and API ergonomics have reasonable defaults an agent can apply. A human review is advisable but not strictly required for the core implementation.