ResearchLLMs Do Not Benefit From Own Prior Responses Context PollutionarXiv·medium signalXBlueskyLinkedInCopy linkRemoving prior assistant responses yields up to 10x context length reduction with minimal quality loss due to context pollutionSourceSource pagearXiv↳ Follow the threadPolicy dependency / Stack layerCodeBLEU Scored 91% for Both RAG Strategies While One of Them Hallucinated APIs 56.4% of the TimearXiv 2609.12464Policy dependency / ContrastAn audit of a production agent pipeline found a reported p99 latency of 2,147,483,647 ms, the signed 32-bit maximum, from a lifecycle clamparXiv 2609.12017Stack layer / ContrastReconstructing Facts From Feed-Forward Residual Vectors Answers Single-Fact Questions in Two-Million-Token ContextsarXiv 2609.12686Stack layer / ContrastRemoving the Tenant ID From an MCP Tool Schema Blocks Cross-Tenant Reads That a Validated Parameter Let Through 26 Times Out of 26arXiv 2609.14780Policy dependency / Stack layerRIPPLE: an edit confined to one prompt-policy segment changes downstream behavior, so replay candidate edits after previously accepted ones before persistingarXiv 2609.12127Stack layer / Follow-up threadPlanting Benign-Sounding Reasoning in an Agent's Context Evades Chain-of-Thought Monitors 25-33% of the TimearXiv 2609.15989Stack layerAn Audit-Grade Agent Run Capsule Replays Every Model Response but Only Completes 2 of 10 Tool-Using WorkloadsarXiv 2609.12582Policy dependency / Stack layerPython's Import Statement Is an Execution Boundary: 90% of Initialization-Activated Advisory Vulnerabilities Are High or CriticalarXiv 2609.14791