FactGate: A Task-Scoped Data Gateway for AI Agents with Cumulative Exposure Accounting

arXiv:2607.00751 · cs.DB, cs.CR · Submitted 2026-07-01 · Read on arXiv

cs.DB, cs.CR

Submitted: 2026-07-01

Updated: 2026-10-11

Comments: 12 pages plus a 37-page supplement. This version supersedes the earlier draft titled "SessionBound" (v1) with a new title, system and evaluation. Code: https://github.com/minmin-lab/factgate . Raw evaluation evidence: https://doi.org/10.5281/zenodo.23149437

Code: https://github.com/minmin-lab/factgate

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• AI Agents — 3× in this paper · explained in 2 episodes

Related papers

• Cost-Aware Mixture-of-Experts Coordination for Model Markets
• Vibe Coding on Trial: Operating Characteristics of Unanimous LLM Juries
• Human-Level Text-to-SQL via Reinforcement Learning on Verified Data, Without Pipeline Engineering
• MetaJoin: Adaptive Join Selections for Heterogeneous Tables
• Bridging Business Intent and Data: A Benchmark for Automatic Relational Data Product Generation
• DexterSQL: Deep Schema Exploration and Rule-based Correction for Text-to-SQL Generation