Project · active

Coding Agent Harness Study (I): Architecture Comparison and Restricted File Execution

Starting from public material, observable behavior, and a runnable microkernel, learn how to turn a coding agent's permissions, tool calls, and execution traces into a verifiable system boundary.

LiyukStarted August 13, 2026August 14, 2026GitHub ↗

A conceptual illustration of a task core surrounded by layered boundaries, feeding an auditable event trace into a restricted file workspace. — Conceptual illustration: capabilities are constrained by permissions, approvals, and traceable execution boundaries; not a screenshot of the project interface.

Maybe related to this one.