Awesome Harness Engineering
A curated, multilingual list of resources, patterns and templates for harness engineering: the scaffolding of context delivery, tool interfaces, planning artifacts, verification loops and memory that surrounds an AI agent.
Awesome Harness Engineering collects resources on harness engineering, the discipline of designing the scaffolding around an AI agent, including context delivery and compaction, tool design, skills and MCP integration, permissions and authorization, memory and state, task orchestration, verification and CI integration, and observability and tracing.
Its foundations section links canonical essays from OpenAI and Anthropic on agent-loop design and long-running task harnesses, and further sections cover reference implementations, including tutorials, generators and demo harnesses, along with security and sandboxing, evaluation and verification, templates, and production infrastructure. The list is deliberately focused on the harness rather than the underlying model, on the premise that harness components exist because a model alone cannot yet do the job, and are expected to become less necessary as models improve.