Open Source
All my open source work in one place. Production tools, research implementations, and coordination infrastructure for multi-agent systems.
Quick Links
| Category | Projects | Link |
|---|---|---|
| Coordination Layer | 4 repos | MCP/A2A extensions, patterns library |
| Core Implementation | 1 repo | HCT reference implementation |
| Developer Tools | 6 repos | AntiSlop, EnvCheck, NerfStatus, SpeakOps |
| Research Infrastructure | 1 repo | HCT benchmark suite |
Coordination Layer: HCT for Multi-Agent Systems
The problem: MCP connects agents to tools. A2A connects agents to each other. But neither tells agents when to act, how urgently, or whether to wait for approval.
The solution: Borrow coordination semantics from musical orchestras. Seven signals that work anywhere.
| Signal | When to Use | Example |
|---|---|---|
| CUE | "Your turn—act now" | Task dispatch |
| FERMATA | "Hold for approval" | Human-in-the-loop |
| ATTACCA | "Immediate handoff" | Real-time flows |
| VAMP | "Loop until ready" | Quality checks |
| CAESURA | "Full stop" | Emergency shutdown |
| TACET | "Stay silent" | Resource conservation |
| DOWNBEAT | "Sync point" | Barrier synchronization |
Quick Start
# MCP Extension
pip install hct-mcp-signals # Python
npm install @hct-mcp/signals # Node.js
cargo add hct-mcp-signals # Rust
go get github.com/stefanwiest/hct-mcp-signals/go # Go
# A2A Extension
git clone https://github.com/stefanwiest/hct-a2a
cd hct-a2a/samples/python && python conductor.py
→ Read the HCT Specification · → View Research
Developer Tools
Production-ready tools for DevSecOps and AI engineering.
| Tool | Purpose | Install |
|---|---|---|
| AntiSlop | Multi-language linter for AI-generated code slop | npm install -g antislop |
| EnvCheck | Environment file linting for DevSecOps | cargo install envcheck |
| NerfStatus | Scientific LLM degradation detection | pip install nerfprobe |
| SpeakOps | Real-time AI speech coaching | Desktop app |
→ Visit Tools page for detailed documentation.
🏗️ All Repositories
Coordination Layer
HCT Spec & RFC
The canonical specification for HCT signals and performance parameters. Source of truth for all implementations.
HCT-MCP Signals
The coordination layer for MCP. Available in Python, TypeScript, Rust, and Go.
HCT-A2A Extension
HCT coordination signals for Google's A2A Protocol. Polyglot samples in Python, TypeScript, Rust, and Go.
Research Infrastructure
HCT Benchmarks
Benchmark suite for HCT coordination patterns. Reproducible performance metrics.
HCT Patterns Library
15+ diagnostic patterns for multi-agent coordination failures. Enterprise-ready solutions.
Developer Tools
AntiSlop
Multi-language linter for detecting AI-generated code slop—placeholders, deferrals, hedging, and empty stubs. Tree-sitte…
EnvCheck
Environment file linting for DevSecOps. Detects .env issues and syncs with Kubernetes, Terraform, Ansible, GitHub Action…
NerfStatus
Scientific LLM degradation detection. Detects silent model collapse, quantization artifacts, and instruction failures wi…
SpeakOps
Real-time AI speech coaching. Analyzes speech during interviews, sales calls, or presentations and sends HUD-style telep…
HCT Core
Reference implementation of the HCT signal bus. Layers 3-5 coordination primitives.