← back to paper
arxiv: 2605.16309 · 2 revisions
ANNEAL: Adapting LLM Agents via Governed Symbolic Patch Learning