Systems, evidence, and what comes next.

My work Case schema and four-way action space, fourteen synthetic cases, the data-driven policy and runtime guard, the evaluation harness, and a client-side port held to the original by a full-grid parity check.

Honest limit The cases and the policy share an author, so zero policy error shows self-consistency, not generalization. The agents are deterministic stand-ins, not language models.

Next Clinician-authored cases held out from the policy author, a second data seed, boundary traps at each threshold, and a real model-driven agent in place of the stand-ins.

My work Backend architecture, LLM integration, prompt contracts, shared memory, World Book retrieval, and decision state.

Next Structured user testing of the public branching demonstration.

My work Rule engine, legal-move model, MCTS opponent, responsive interaction design, accessibility, and deployment.

Next Export the trained policy/value checkpoint for browser inference and compare it against the search-only bot.

Finding The binding constraint was how the budget was allocated, not how large it was. Holding back the characters the rendezvous would need, instead of spending them first-come on map facts, moved the working threshold from 32 characters down to about 12.

Next Price patrol danger as a cost that scales with the remaining clock rather than treating it as impassable; the residual failures are agents orbiting a guard with an unspent risk budget.

Finding Prediction checks are useful as a dense feature, but too brittle to serve as the whole reward.

Next Validate goal-conditioned value differences on held-out trajectory prefixes.

Design choice Local Qwen summarization is the default; paid APIs remain optional.

Next Add citation-aware chaptering and an evaluation set for summary fidelity.