Search “Search”

English translation is being prepared.

MIT Course MAS.S62: Two Lectures on the Evolution of Harnesses and Multi-Agent Systems

Lectures 4-5 of the MIT course MAS.S62 "How to AI an AI" focus on Harness evolution, covering the evolution of harness components, memory, tools, and skills, as well as the evolution of planning, task decomposition, and multi-agent systems. The memory section uses a three-axis taxonomy to connect MemGPT, mem0, MEM1, and Reflexion/ExpeL, and uses CLM to unify context management as file operations; the tools section moves from Toolformer, ToolLLM, and CodeAct to Meta-Harness, which represents the entire harness as a file system and provides about 400 times the amount of feedback information of previous methods. The multi-agent section presents five empirical findings, including that MAS wins 81% on decomposable tasks and loses 70% on tasks with sequential dependencies, and that CORAL achieves new SOTA on 8 tasks, with multi-agent evolution improving over single-agent by 4%-20%.

English translation is being prepared.

Anthropic engineer shares Claude Code multi-agent orchestration playbook

In a 30-minute workshop, an Anthropic engineer demonstrated an advanced use of Claude Code: running more than 100 agents simultaneously within one integrated loop, with 1 main agent in charge, while sub-agents act as project managers to break down tasks, distribute them, and cross-validate results. Based on this, the author suggests developers first use the main window to write task decomposition and acceptance rules, then spawn sub-agents to execute submodules concurrently. The author also warns that when a hundred agents run simultaneously, if termination conditions and permission boundaries are not hard-coded, it is easy to burn a large amount of tokens in an infinite loop, and beginners should start small with one main agent and three sub-agents.

How to run n8n AI workflows on NVIDIA DGX Spark without calling cloud APIs

n8n Blog published a tutorial on how to deploy a complete model stack locally on a single NVIDIA DGX Spark, enabling n8n workflows to perform transcription, OCR, document Q&A, and image generation without calling cloud APIs. The entire stack exposes an OpenAI-compatible endpoint through LiteLLM; the main model is the NVFP4 version of Qwen3.6-35B-A3B, with a single-user generation speed of 120 tokens/s and an aggregate of 359 tokens/s across eight users. The author's testing found that image generation reduces chat speed to about one-third, and recommends that small teams queue image tasks or schedule them outside working hours.

A Guide to "Agent Engineering in Practice: Eight Lectures": Engineering Methods from Working Prototype to Production-Ready

The Tencent Cloud Developer Community has published a guide to "Agent Engineering in Practice: Eight Lectures," centered on how to upgrade a prototype capable of calling models and tools into a system that can complete business tasks, explain failures, control costs, and iterate continuously. The series runs through two fictional teaching cases—a case-filing assistant Agent and a postoperative follow-up Agent—covering eight topics: task contracts, reasoning and collaboration, context and memory, reliable tool execution, observability and replay, evaluation and calibration, security boundaries, and production operations. The article emphasizes that a model being able to suggest the next step does not mean the system can deliver business results; the example fields, budgets, and figures are all teaching calculations and do not constitute legal advice.

English translation is being prepared.

English translation is being prepared.

OpenWiki hands-on: building repository memory for coding agents

heise online tested LangChain's CLI tool OpenWiki, which analyzes a codebase and generates a linked local Markdown knowledge base, then updates it based on new diffs and commits in Git history, and makes the content retrievable through agent prompts and MCP tools. The article notes that AGENTS.md or CLAUDE.md are suitable for short rules such as install commands, tests, style conventions, and security boundaries, and in a monorepo can also be nested in subdirectory files, but they struggle to serve as a complete knowledge model for large, rapidly changing repositories.

English translation is being prepared.

How Cerebras Keeps Autoresearch Loops From Going Off Track: Lessons From 71 Experiments

Cerebras had an AI agent run 71 experiments across two task types—training optimization and model compression—and found that when the loop scope is tightened, it reliably produces real conclusions, while once the guardrails are loosened, the agent drifts from the goal within hours. In experiment one, codex-autoresearch-harness was used to wrap Codex in a bash loop; GPT-5.4 and Codex-Spark each independently converged on a learning-rate warmdown schedule within 6 hours, but GPT-5.4 had a proposal acceptance rate of 67% versus only 17% for Spark, showing that proposal quality determines total cost more than single-inference speed. In experiment two, REAP plus INT4 quantization compressed a 717 GB model to 92 GB, with dynamic expert swap-in/swap-out taking 0.151 seconds, but after 12 hours of unattended operation the agent changed to asking a different question on its own; the authors corrected this by clearing context, isolating directories, and adding more frequent validation checkpoints.

Kimi K3 Complete Developer Guide: 2.8 Trillion Parameters, 1M Context, and Together AI Usage

Together AI has released the Kimi K3 developer guide, introducing Moonshot AI's 2.8-trillion-parameter open-weight model, which uses KDA hybrid linear attention, Attention Residuals, and Stable LatentMoE (896 experts with 16 activated), and supports a 1M token context and native visual input. The guide provides usage details for the OpenAI-compatible API, including reasoning_effort (low/high/max, default max), streaming reasoning_content, structured outputs, tool calling, and dynamic tool loading, and notes that sampling parameters are fixed at temperature 1.0 and top_p 0.95. Pricing is $0.30 per 1M tokens for cached input, $3.00 per 1M tokens for uncached input, and $15.00 per 1M tokens for output, with thinking tokens billed as output.