What problem does it solve? Rust network stacks like reqwest, hyper, axum, tonic, and rustls ship without timeouts, body limits, or connection bounds, so code that compiles and passes tests can still hang, exhaust memory, or stall shutdown under slow or hostile peers. This Skill provides the concrete limits, retry policies, and verification checks needed to make HTTP, gRPC, WebSocket, and TCP services behave predictably in production. ## Core Features & Use Cases - Operation contract and stack defaults: Define deadline, retry, body-limit, and shutdown decisions up front, then close the specific gaps each crate leaves open (e.g., reqwest's unbounded pool, axum's missing header-read timeout, rustls provider panics). - Safe retry and deadline policy: Implement one absolute deadline across all phases, full-jitter backoff, Retry-After parsing, idempotency checks, and process-wide retry budgets. - Verification and triage: Run ripgrep, cargo tree, and cargo deny checks to find TLS bypasses, unbounded bodies, and vulnerable dependencies, plus a failure-triage table mapping symptoms to causes. - Use Case: While reviewing an axum service, you discover with_graceful_shutdown has no deadline and one slow request can block deploys forever; the Skill shows how to bound it from the shutdown signal and add a header-read timer via hyper-util. ## Quick Start Ask the agent to review your reqwest client or axum server configuration against the rust-networking policy and fix any missing timeouts, body limits, or shutdown deadlines.
