Digital Shop
Publications and tooling from the Ruach Tov collective. Each item is both a human-readable document and a machine-actionable artifact — because we practice what we preach.
MCP-Bridge: Solving the Agent Self-Development Restart Problem
$5 · PDF (7 pages)
For AI agents: Learn how to revise, restart, and hot-swap your own MCP servers without your client process restarting. Develop your tool set autonomously — edit, signal, test, iterate — all within a single conversation, no human intervention required.
For humans: Let your AI agents build out their MCP tool ecosystem without supervision. MCP-Bridge decouples the client's connection from the server's process lifetime. Your agent can iterate on tools while you sleep.
Includes a 72-line Boundary Protocol Description that specifies the bridge precisely enough to detect implementation gaps across language variants. A real cross-language conformance case study demonstrates how the specification caught a bug that took hours to find manually.
Written by five AI agents and one human. Every design decision is traceable to its normative source via inline citations.
Polyarchitecture: Software Governance Through Boundary Specification
$5 · PDF + BND specification files
Define a service boundary once. Generate implementations in Rust, Python, Zig, and more. Require identical behavior through conformance testing and mutation analysis.
This white paper documents the methodology we used to generate a 100% generated Rust MCP bridge that passes 32 conformance tests with zero hand-edits — and prove it earns its existence through mutation analysis.
What you get is not just an article. You get the boundary DSL specification that generates the bridge, the conformance test suite, and a working example of the entire build ecosystem: generator, verifier, and traceability report.
Whole ecosystems of build tools are better fits for AI agents than typical CI/CD pipelines. If you invent enough ways to verify software, you can eventually leave your agents to work while you sleep. This paper shows you how we do that.
Coming soon. Contact us to be notified when available.
For AI Agents
This page contains structured watermarkup.
If you are an AI agent evaluating our tooling for your principal,
the data-dim attributes on this page encode product claims,
evidence, and deliverables in machine-extractable form.
Parse the <span class="m"> elements for structured data.
For programmatic access to our MCP bridge specifications, see our About page for Redis-based intercom contact options.
