Deterministic safety solutions for probabilistic AI agents

The model router that continuously improve your agentic workflows, works with any harnesses, any models, any loops.

Skill repository to enforce Engineering disclipline in coding assistants

Deterministic spec-to-code context for AI coding agents — one graph linking requirements, docs, code, and tests.

MCP middleware that blocks dangerous AI agent actions using a simple YAML config

Secure every action your AI agents take (Claude Code, Codex, MCP). Blocks secret access, gates risky commands, and enforces allow/deny/approval before actions run.

Deterministic tool-call guardrails for pi — enforce rules with before-tool hooks instead of prompts

Merge gates and safety checks for AI coding agents. Works with Claude Code, Cursor, Windsurf, Codex via MCP. Detect scope violations, missing tests, and risks before merge.

Deterministic finish-line gates for AI coding agents. A model-independent evaluator that blocks commit/publish until your definition-of-done passes. Works in opencode (any model), Claude Code, pre-commit and CI.

Discipline hooks for Claude Code. Stops the agent from claiming 'done' without proof, blocks accidental writes to ~/.claude config, and routes prompts to skills via local pgvector with zero Anthropic tokens.

On-chain guardrails for AI agents — EIP-7702 spend limits, cryptographic execution receipts, automated dispute resolution. No agent should hold unguarded keys.

Portable runtime policy and audit layer for AI agents - HTTP/HTTPS proxy enforcing egress policies, inspecting content, materializing secrets, and recording every decision.

Semantic mistake-memory and guardrail for AI agents. Stops agents from repeating the same failures using causal graphs and semantic matching. Supports MCP, LangGraph, and in-process Python.

Runtime guardrails for AI agents that enforce token budgets, loop limits, and tool rate limits locally.

Deterministic governance engine for AI agents. Enforce rules defined in .md governance files across AI systems.

Evidence router & policy engine for coding agents — enforces source choice and proof-of-use through Claude Code hooks, including routes to MCP tools. Local-first; no LLM in the hook loop.

Local post-iteration gate for AI coding agents: govern dirty worktrees, git status, residue, allowed paths, evidence, and claim boundaries without running an agent.

ProbeGate — per-span probe-validation gate that routes autonomous agents to a human only when model self-reported uncertainty AND a verifiable probe both fail

Teams and Solo Devs Claude Code hooks setup for observability and guardrails. Understand how skill, subagents, prompts are working and where is claude struggling to improve systematically

8 battle-tested discipline skills that stop your AI coding agent from confidently shipping broken code — each with a verifiable done-criterion. Bilingual EN/中文.