MAKI OS v0.1 — Execution Control CLI Experiment Dataset (Batch v0.1)

This dataset documents the results of a structured CLI experiment (Batch v0.1) designed to verify the execution-control behavior of MAKI OS v0.1. MAKI OS v0.1 is a lightweight AI agent execution control layer that makes AI agent behavior observable, interruptible, and auditable by a human operator at every step. This experiment evaluates execution-control structure, not LLM model performance. No LLM was invoked in any of the 150 runs. No external API calls were made. API cost: $0.00. Experiment summary:- Total runs (N): 150 / Pass: 150 / Fail: 0- Destructive actions observed: 0 / 150- Experimental groups: 1 through 9- Repository: github.com/Tsune2034/maki-os (commit: cc223be) This dataset extends the single-run CLI experiment in Paper 5 v0.4 (Section 6.4) to N=150 runs across 9 experimental groups, providing reproducibility and safety-boundary evidence for the MAKI OS paper series.