harness-creator-v3
Harness Creator v3
Use this skill to make a repository easier for coding agents to start, stay in scope, verify work, and resume across sessions. Keep the harness small enough that agents actually follow it.
New in v3: Embedded behavioral policies. The generated AGENTS.md now includes a Coding Policy (the Ponytail ladder), Coding Standards, Editing Discipline (surgical changes), test-first Definition of Done, proactive assumption surfacing, and safety carve-outs. Validation scores these behavioral subsystems alongside the five structural subsystems. See CHANGELOG.md for the full list.
Not for model selection, prompt tuning in isolation, chat UI design, or general app architecture.
Core Model
Every useful coding-agent harness has five structural subsystems plus behavioral policies: