Repair Async Generator Cleanup Bugs

The problem

Async generators often leak resources or keep producers running long after a consumer exits or an early break occurs. Improperly handled cancellation signals and races between pending next calls and disposal lead to flaky tests and memory leaks.

What it does

• Identifies orphaned producer tasks and unreleased resource leases in async iteration flows.
• Enforces a canonical termination contract across early returns, throws, and timeouts.
• Wires AbortSignal cancellation through nested adapters to ensure it reaches the producer.
• Ensures idempotent cleanup that preserves the original error instead of masking it.
• Implements deterministic test barriers to replace unreliable wall-clock sleeps.

Frameworks & tools

Works with JavaScript/TypeScript async iterators, Node.js streams, AbortController, and repository-level concurrency primitives.

Why this beats prompting it yourself

Generic prompts often miss the nuances of the async event loop, such as settling pending promises during disposal. This skill uses a formal Resource Ownership Register to trace every generator generation, preventing the "swallowed error" and "double cleanup" bugs common in manual patches.

Use cases

• Fixing database cursors that remain open after a for-await...of loop breaks.
• Propagating cancellation from a UI component down to a background data stream.
• Repairing iterator adapters that accidentally convert a throw into a normal completion.
• Eliminating race conditions where a producer publishes a value after the consumer disconnected.

Known limitations

Does not change payload ordering, retry policies, or protocol semantics. It cannot contact live production brokers or databases and relies on local fakes for validation.