Specialized skill for designing and bootstrapping **effective harnesses** for
long-running AI agents.
Combines Anthropic's long-running agent scaffolding, AIHero's
Plan-Execute-Test-Commit loop, and Revfactory Harness patterns.
Use this skill whenever you need agents to maintain state, make incremental
progress, and leave production-ready code across multiple sessions.
Harness-layer skill with clean stratified disclosure. Fully portable across agentskills.io environments and models when paired with a compatible harness. Always activate together with `code-writer` + the disclosed book (and domain skills the harness discloses).
