Best Guardrail AI Skills & MCP Servers

Your AI agent has read your codebase but still can't safely change it. unerr is a local guardrail that hands the agent the call graph and your rules at the moment it edits.

CLI and MCP servers for Cutline, including SlopBurn: a product quality engineering roguelike RPG for vibecoding workflows.

MCP control plane for local cognitive labor — job-shaped tools with tiered Ollama models (instant/workhorse/deep/embed), server-enforced guardrails, and measured economics so Claude can delegate bulk work without losing control.

CLI and MCP servers for Cutline, including SlopBurn: a product quality engineering roguelike RPG for vibecoding workflows.

Production guardrails for AI coding agents

The runtime security layer for AI agent commerce. Drop-in CLI + MCP server — blocks hallucinated purchases and keeps card credentials out of agent context. It only takes 0.1% of hallucination to drain 100% of your wallet.

MVA (Model-View-Agent) framework for the Model Context Protocol. Structured perception packages with Presenters, cognitive guardrails, self-healing errors, action consolidation, and tRPC-style type safety — so AI agents perceive and act on your data deter

A TypeScript library for selective context access, visibility control & safety enforcement for LLMs

Security, cost, and health governance proxy for MCP infrastructure — three-layer detection engine (regex + schema + LLM), monorepo, corpus, CI/CD

MCP server that proxies all agent actions, classifies risk, enforces natural-language policies, and logs to a local SQLite audit DB with a live dashboard.

CLI and MCP servers for Cutline — authenticate, then run constraint-aware MCP servers in Cursor or any MCP client.

BLACK_WALL MCP server — a pre-action risk check your AI agent calls before any irreversible action (send email, move money, run SQL, delete data).