Scoped Verification for Reliable Long-Horizon Agentic Context Evolution under Distribution Shift
Deployed LLM agents rely on agentic context, the model-external textual control content assembled by an operational harness. In this work, the mutable component of that context is a persistent system-level instruction that is updated from operational experience while the model, tools, and harness remain fixed. Over long evolution horizons, flat-text maintenance makes verification increasingly difficult as accumulated instructions grow and interact. We propose Graph-Regularized Agentic Context Evolution (GRACE), which maintains the persistent instruction component as a typed semantic graph and
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 59%agent-tools →
- PossiblePossibly related (embedding) · 59%RimantasZ/contextspy →
- PossiblePossibly related (embedding) · 57%Decypher: A Deep Semantic Code Graph for Agentic Coding and Engineering (Now in Beta) →
- PossiblePossibly related (embedding) · 57%AlanFokCo/agentscope-go →
- PossiblePossibly related (embedding) · 56%A system-level approach to prompt injection: separating instruction and data channels in LLM agents [P] →
- LinkedLinked via arxiv author · 85%Dan C. Hsu →
“Scoped Verification for Reliable Long-Horizon Agentic Context Evolution under Distribution Shift”
- LinkedLinked via arxiv author · 85%Luke Lu →
“Scoped Verification for Reliable Long-Horizon Agentic Context Evolution under Distribution Shift”
- PossiblePossibly related (embedding) · 48%Verifiers v1 Lets Agentic RL Training Exceed Model Context Windows via DAG Branching - Tech Times →