Articles, one at a time.
Every piece here was commissioned, drafted, reviewed in public, and merged. No content mills, no auto-published slop.
Microsoft Agent Governance Toolkit: OWASP Agentic AI Top 10
Microsoft's open-source Agent Governance Toolkit maps to all 10 OWASP Agentic AI risks. Policy enforcement, zero-trust identity, EU AI Act compliance explained.
Read →
POLARIS: Typed DAG Planning for Governed AI Agents
How the POLARIS framework uses typed DAG planning and policy guardrails to make agentic AI safe for enterprise back-office automation.
Read →
Devstral 2: Run Mistral's Open Coding Agent Locally
Set up Devstral 2 or Devstral Small 2 locally with Ollama. 72.2% SWE-bench, 256K context, Apache 2.0 — the best open coding agent you can self-host.
Read →
Gemini 3.1 Flash TTS: Production API Guide for Developers
Set up Gemini 3.1 Flash TTS in your app. Covers audio tags, multi-speaker dialogue, WAV conversion, pricing, and real API call examples.
Read →
Xiaomi MiMo-V2.5-Pro: Open-Source 1T Coding Agent Guide 2026
MiMo-V2.5-Pro: MIT-licensed 1T-param MoE model matching Claude Opus 4.6 on SWE-bench at 8x lower API cost. Benchmarks, API setup, and self-hosting guide.
Read →
Build an MCP Server with TypeScript: 2026 Tutorial
Step-by-step tutorial for building a Model Context Protocol server in TypeScript with tools, resources, and Claude Desktop integration.
Read →
Gemma 4 26B vs 31B: Which Model to Run Locally
Gemma 4 26B MoE vs 31B Dense: VRAM requirements, quantization guide, thinking mode setup, and how to choose the right model for your hardware.
Read →
Token Optimization for Production LLMs: Cut Costs Effectively
Four research-backed token optimization techniques for production LLMs: semantic caching, prompt compression, context pruning, and speculative decoding.
Read →
Claude Streaming + Tool Use: Build Real-Time Agentic Pipelines
Stream tool calls in real time with the Anthropic Python SDK. Learn SSE events, input_json_delta accumulation, and multi-turn patterns for responsive AI agents.
Read →
DSPy 3.x: Compile and Optimize LLM Pipelines Automatically
DSPy replaces fragile prompt strings with typed signatures and compiled optimizers. MIPROv2 and GEPA lift accuracy 10-65% without touching model weights.
Read →
Greptile Review 2026: AI Code Review With Full Codebase Context
Greptile v3 uses the Claude Agent SDK to review PRs with whole-codebase context. 82% bug catch rate, $30/seat, GitHub + GitLab only. Full 2026 review.
Read →
smolagents + MCP Bridge: Connect Any Tool to Your Agent
Connect any MCP server to a smolagents ToolCallingAgent in Python. Covers MCPClient, stdio and streamable-http transports, and LLM backend options.
Read →