← 研踪 PaperTrailarXiv:2611.03434 · v1
Optimize the Harness, Not the Model: Fitting a Minimal Agent to a Self-Hostable LLM for Investigative Analysis
首次提交 2026-11-03 00:53
版本更新 2026-11-03 00:53
雷达观测 2026-11-04 23:41
摘要
Government intelligence teams need LLM agents that can analyze large, messy evidence collections, but security and cost constraints often rule out hosted frontier models and broad, hard-to-audit tool surfaces. We report CLEO: Constrained Loop for Environment Optimization, a method for fitting a minimal agent harness to a fixed, self-hostable LLM. Given an expert-validated dataset, CLEO admits only the tools, guardrails, memory policies, and loop components that improve accuracy while remaining general and auditable. On our held-out dataset, the CLEO harness reaches an 83.8% pass rate, within 1.2 points of the strongest out-of-the-box Claude Code SDK baseline, while being two orders of magnitude smaller.
作者备注：3 pages, 1 table. Industry Day talk at CIKM 2026, Rome, 7 November 2026
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-03 00:53 · 最新观测版本
Optimize the Harness, Not the Model: Fitting a Minimal Agent to a Self-Hostable LLM for Investigative Analysis
Government intelligence teams need LLM agents that can analyze large, messy evidence collections, but security and cost constraints often rule out hosted frontier models and broad, hard-to-audit tool surfaces. We report CLEO: Constrained Loop for Environment Optimization, a method for fitting a minimal agent harness to a fixed, self-hostable LLM. Given an expert-validated dataset, CLEO admits only the tools, guardrails, memory policies, and loop components that improve accuracy while remaining general and auditable. On our held-out dataset, the CLEO harness reaches an 83.8% pass rate, within 1.2 points of the strongest out-of-the-box Claude Code SDK baseline, while being two orders of magnitude smaller.
查看此版本解读 →查看此版本原文 ↗