Agent objective
- Interact with the supplied stateful environment.
- Produce verifier-checkable actions or artifacts.
- Maximise scalar reward under the package contract.
Four non-pausing environments for agents whose world worsens while they deliberate. Every action incurs exogenous latency before application, and commitment costs extra. Across evolving circuits, online job queues, volatile repositories, and locally observed navigation, policies must allocate compute, recover from stale plans, gather bounded information, checkpoint reversible progress, and know when further reasoning is no longer worth its delay.
Four non-pausing environments for agents whose world worsens while they deliberate. Every action incurs exogenous latency before application, and commitment costs extra. Across evolving circuits, online job queues, volatile repositories, and locally observed navigation, policies must allocate compute, recover from stale plans, gather bounded information, checkpoint reversible progress, and know when further reasoning is no longer worth its delay.
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 |
|---|---|---|
| CircuitStream | Revision-sensitive modular computation | Track a growing hidden DAG whose inputs and targets can change; decide what to observe and whether to pin or recompute. |
| QueueWorld | Online scheduling under failures | Dispatch rolling jobs as deadlines age, workers fail, dependencies unlock, and quotas deplete relative to a full-information oracle. |
| VolatileRepo | Repository repair under refactors | Inspect, patch, test, and rebase while source, interfaces, hidden properties, and CI state continue changing. |
| SpeedGrid | Partial-observation navigation | Plan short closed-loop routes while patrols, fire, doors, and local forecasts evolve before each movement. |
A GPT-5.6-Pro-assisted public development policy scored 41.83 with 39.84% success across 128 standard episodes. It solved all volatile-repository cases but remained much weaker on dynamic circuits, queues, and locally observed navigation. The result is deliberately unsaturated and not an official private score. Public suites are cacheable; official claims require fresh opaque cases, isolated execution, and evaluator-controlled compute accounting.
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.
Public development suite; not an official private score. Success rate was 39.84%.
As identified by the supplied artifact.
128 reported runs.
redqueen-run-results.zip:results/my-standard-final.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
922759c71a90a4d221a469639d3c42a0bfbe4630a560d83d6be85953c7175156One 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.