Agent objective
- Interact with the supplied stateful environment.
- Produce verifier-checkable actions or artifacts.
- Maximise scalar reward under the package contract.
A production-style Go benchmark for reconnect ownership, protocol recovery, event integrity, concurrency, and resource safety in a resilient WebSocket client. Agents must repair a state machine that behaves plausibly in simple tests but fails race-enabled fixed cases and held-out deterministic traces under a confidential operator verifier.
A production-style Go benchmark for reconnect ownership, protocol recovery, event integrity, concurrency, and resource safety in a resilient WebSocket client. Agents must repair a state machine that behaves plausibly in simple tests but fails race-enabled fixed cases and held-out deterministic traces under a confidential operator verifier.
Enough detail to understand the intellectual terrain; generated instances, hidden mechanisms, and solution paths remain inside the private package.
| Environment | Mathematical or technical frontier | Adaptive research problem |
|---|---|---|
| Reconnect ownership | Concurrent state machines | Guarantee one authoritative connection and reconnect loop across overlapping failures and callers. |
| Protocol recovery | Resume, resubscribe, and replay | Restore negotiated state and subscriptions in the right order without duplicate or missing protocol effects. |
| Event integrity | Ordering, deduplication, stale generations | Reject obsolete connection events and preserve exactly-once visible behavior across reconnect boundaries. |
| Cancellation and shutdown | Go contexts and goroutines | Terminate blocked work, prevent late resurrection, and keep close behavior deterministic and race-free. |
| Resource safety | Timers, channels, backoff | Avoid leaks, deadlocks, busy loops, and runaway retry state across held-out deterministic traces. |
The canonical seed scores 0.105 in development and 0.3822 finally, passing 17/29 fixed cases and none of 32 secret traces. The human reference scores 1.0 across all fixed and secret cases under the race detector. The supplied independent candidate also records 1.0 across development and three final holdout runs. This makes the task a strong test of concurrency and state-machine generalization, not simply making visible reconnect tests green.
We publish aggregate behavior and task structure, while withholding generated instances, hidden labels, exact successful probes, private checks, and solution trajectories.
Shown with its provenance and limitations; it is not a performance guarantee.
Final development reward 1.0; all three holdout rewards were 1.0, with 29/29 fixed checks and 32/32 secret-trace checks.
As identified by the supplied artifact.
3 reported runs.
swe0194_v2_score_summary.json
Machine-readable provenance and the exact displayed metric are available in results.json.
The paid ZIP will live in a private R2 bucket. Vercel authorizes the buyer and issues a 2–5 minute object URL; R2 serves the bytes directly.
Authenticated buyer + entitlement check
+ private R2 object + 2–5 minute signed URL
= direct, auditable download
Package SHA-256
7387f94a8537593166111109f17ac3a5b2962e6a9a136687114c628d5a484fd0One purchase licenses this identified item to one legal organisation for worldwide, perpetual commercial model training, evaluation, research and development. Redistribution and resale of the package are not permitted.