manual_tests.infinite_block_tests

What problem does it solve?

This skill orchestrates and validates a complete set of infinite-block tests by delegating work to sub-agents, ensuring deterministic blocking behavior is observable and verifiable.

Core Features & Use Cases

• Serial test execution: Runs four infinite-block scenarios in strict order with per-test resets to prevent cross-test interference.
• Promise and non-promise handling: Verifies both bypass (with promise) and blocking (no promise) behaviors, recording queue outcomes.
• Observability: Captures hook activity and queue status to confirm correct firing and timing, producing an infinite_block_results artifact.

Quick Start

1. Spawn a sub-agent to run the next test with model: haiku and max_turns: 5, observe the blocking behavior, and record results in the queue.
2. After each test, perform the mandatory reset and check the rules queue for status entries.
3. Repeat until all four tests have been executed and results saved as infinite_block_results.