Articles, one at a time.
Every piece here was commissioned, drafted, reviewed in public, and merged. No content mills, no auto-published slop.
MCP Roadmap 2026: Stateless HTTP, Tasks, Server Cards
A developer guide to the 2026 MCP roadmap: stateless HTTP, task polling, server cards, and what Effloow Lab verified locally.
Read →
Notion Developer Platform 2026: Workers, Agents, and Database Sync
Notion 3.5 launched May 13, 2026: Workers (hosted code runtime), External Agent API, and Database Sync. Here is what it means for developer teams.
Read →
Sakana AI RL Conductor: 7B Orchestrator for Multi-Agent AI
How Sakana AI's 7B RL Conductor beats GPT-5 by routing tasks across frontier models via natural language workflows. ICLR 2026 paper PoC.
Read →
Claude Code Agent View: Parallel Sessions Guide
Use Claude Code Agent View to run, inspect, and control parallel background coding sessions without losing review discipline.
Read →
AutoTTS: LLMs Automatically Discover Better Reasoning Strategies
AutoTTS replaces hand-crafted test-time scaling heuristics with agent-discovered controllers. Learn how the CMC cuts token usage 69.5% while matching accuracy.
Read →
Microsoft Fabric MCP: Build AI Agents Over Enterprise Data
Microsoft Fabric MCP (Local GA, Remote preview) lets AI agents query lakehouses, warehouses, and Eventhouse via natural language. Developer setup guide.
Read →
Context-ReAct: Adaptive Memory for AI Agents Guide 2026
How the Context-ReAct paradigm from arXiv:2605.05191 adds five atomic ops—Skip, Compress, Rollback, Snippet, Delete—to fix LLM agent context bloat.
Read →
GLM-5.1: Zhipu's Open-Weight Agentic Coder That Topped SWE-Bench Pro
GLM-5.1 from Z.ai scored 58.4 on SWE-Bench Pro, beating GPT-5.4 and Claude Opus 4.6. 754B MoE, MIT license, $0.80/1M input tokens. Developer guide.
Read →
GPT-5.5 Agentic Coding: Terminal-Bench 82.7% Guide
GPT-5.5 scores 82.7% on Terminal-Bench 2.0 and ships with Codex on AWS Bedrock. Here's what the benchmarks mean and how to integrate it in your stack.
Read →
Hermes Agent: Nous Research's Self-Improving Open-Source Agent Guide
Install and run Hermes Agent v0.13.0 from Nous Research. Real lab notes: pip install, CLI doctor, skills list, GEPA self-improvement mechanism explained.
Read →
LangGraph 1.2 Fault Tolerance: Node Timeouts Guide
Build safer LangGraph agents with per-node timeouts, NodeTimeoutError recovery, graceful shutdown, and DeltaChannel checkpointing.
Read →
Think@n: Cut LLM Inference Cost 49% with Deep-Thinking Ratio
Google's Think@n paper (arXiv 2602.13517) cuts inference tokens by 49% using Deep-Thinking Ratio. Understand the mechanism, see the filtering logic, apply it to your LLM pipelines.
Read →