Agent Harness Engineering — References

Sources inspected for the harness boundary and concrete environment designs; reviewed 2026-10-13.

• Agent Harness Engineering — Addy Osmani. A practitioner framing of the software surrounding a model; useful terminology, not a normative architecture specification.
• Effective harnesses for long-running agents — Anthropic, November 26, 2025. A concrete initializer/progress/verification approach across fresh context windows. The experiment's task scope limits generalization.
• Harness engineering: leveraging Codex in an agent-first world — OpenAI, February 11, 2026. Repository knowledge, executable constraints, and inspectable feedback as parts of the engineering environment; an experience report rather than a benchmark.
• Deep Agents overview — LangChain. Maintained documentation for a packaged agent harness; check version-specific defaults.

The component map and diagnostic sequence in this capability are an engineering synthesis of these concerns. They do not claim a standardized definition of every harness boundary.