MemoHarness: Optimizing the Agent Harness
omarsar0 · x · 2026-07-17
This paper focuses on optimizing the agent harness—the external control layer that turns a base LLM into an executable agent.
The core method, MemoHarness, decomposes the harness along the inference process into 6 editable control planes: context, tools, generation, orchestration, memory, and output. Without relying on test-time labels, feedback, gradient updates, or additional search, the system adapts to new cases by retrieving similar historical cases based on diagnostics and cross-case patterns.
It achieves a score of 0.806 on the shell-agent benchmark, outperforming the strongest fixed harness baseline at 0.722, while maintaining a lower per-task cost than the most powerful commercial baseline compared in the text.
More from coding & agent
- Alchemy adds effect/Workflow support across Durable Objects, Workflows and Lambda Durable Functions — samgoodwin89 · 2026-10-27
- Alchemy's Lambda Web Functions run plain Node web servers on Lambda, no OpenNext needed — samgoodwin89 · 2026-10-27
- Alchemy Domain brings domain-driven design and story testing to Effect backends — samgoodwin89 · 2026-10-27
- Alchemy makes all ~2,850 AWS bindings work from any cloud, not just S3/Lambda/DynamoDB — samgoodwin89 · 2026-10-27
- Production AI agent scored 95% robustness at baseline, 37% under adversarial attack — njyx · 2026-10-27
- Chaining Grok Bot and Codex to detect and mass-block AI reply accounts — vista8 · 2026-10-27