BUILD
DIFFERENT
AI AGENTS.

For developers who build AI systems the hard way — local inference, MCP servers, multi-agent orchestration, and agent memory. No tutorials recycled from the official docs. Real patterns from production builds.

// Latest from the Feed
MCP
Building a multi-tool MCP server with FastMCP in 30 minutes
12 min · March 2026
Agents
Fan-out orchestration: running 10 agents in parallel against a single task
9 min · March 2026
Local LLM
Qwen-Coder-30B on consumer hardware: real benchmarks, real limits
7 min · Feb 2026
Architecture
NATS JetStream for agent message queuing: why it beats Redis Streams
14 min · Feb 2026
34
Published Posts
12
OSS Repos
100%
Production-Tested
0
Recycled Tutorials
Posts

READ THE REAL STUFF.

The Right Way to Build Multi-Tool MCP Servers (And the Three Most Common Mistakes)

After building 25+ MCP servers, here are the patterns that hold up in production and the shortcuts that blow up in your face when an agent starts hammering the endpoint.

Running Qwen-Coder-30B on Dual RTX 3060s: Setup, Benchmarks, Limits

Real numbers, not marketing copy. What it can and can't do on consumer hardware.

7 Types of Agent Memory (And When to Use Each One)

Episodic, semantic, procedural — not just theory. Built all 7, here's what each one actually changes.

Fan-Out/Fan-In: Running 10 Agents in Parallel with ThreadPoolExecutor

NATS JetStream vs. Redis Streams for Agent Message Queuing

Refactoring a 100x Coding Agent: 7 Modules, 34 Tests, Zero Regressions

All 34 Posts →
Open Source

TAKE THE CODE.

Everything we build that isn't a client product goes public. MIT license. Fork it, improve it, ship it.

agent-memory-server
7-type persistent agent memory with ChromaDB + SQLite. TF-IDF retrieval. Plug into any MCP stack.
Python★ 142→ GitHub
parallel-orchestrator
Fan-out/fan-in multi-agent execution skill. ThreadPoolExecutor + task routing. Production-hardened.
Python★ 89→ GitHub
mcp-server-template
FastMCP boilerplate with auth, rate limiting, error handling, and structured logging. Start here.
Python★ 201→ GitHub
lm-studio-agent-loop
Production MCP client agent loop for LM Studio. Multi-tool calling, retry logic, token tracking.
Python★ 67→ GitHub
freecad-mcp
38-tool MCP server for headless FreeCAD. Architectural floor plan drafting from natural language.
Python★ 44→ GitHub
token-tracker-dashboard
Real-time token usage dashboard for multi-agent systems. TikTokenTracker + Structlog JSON backend.
JavaScript★ 38→ GitHub
All 12 Repos →
// Weekly

THE REBEL
DEV FEED.

One email a week. One deep-dive on a real pattern from a real build. MCP, agents, local LLM, orchestration. No beginner tutorials. No startup hype. Just the actual work.

// No spam. Unsubscribe anytime. MIT license on all code samples.
Membership

ALL SIGNAL.
NO NOISE.

Free Reader
$0
forever · no card
All posts, all OSS repos, weekly newsletter. Always free.
  • All published posts
  • All OSS repositories
  • Weekly newsletter
  • Community Discord
Start Reading
Build With Me
Custom
via Aether Neural
Hire the person who builds all this for your actual production system.
  • Custom MCP server builds
  • Multi-agent system design
  • LM Studio + local LLM setup
  • Monthly retainer available
→ Aether Neural
Docs

START HERE.

MCP Server in 5 Minutes

This is the minimal FastMCP server template we start every build with. Copy it, rename the tools, ship it.

# pip install fastmcp from fastmcp import FastMCP mcp = FastMCP("my-skill-server") @mcp.tool() async def do_thing(input: str) -> str: """Does the thing. Describe it here — agents read this.""" return f"Did: {input}" if __name__ == "__main__": mcp.run()
Full Docs →