ai-safety

The action firewall for AI agents. Enforce policy and human approval before risky tool calls, shell commands, workflows, and production changes, with auditable evidence.

An orchestration runtime for multi-agent AI systems. Declare agents, tools, and policies as YAML; Orloj schedules, executes, routes, and governs them for production-grade operation.

The vendor-neutral protocol for AI agent safety & security — and the neutral benchmark that ranks the vendors.

A lightweight, highly secure AI API Gateway/Proxy written in Go. Acts as transparent middleware between local AI coding clients (OpenCode/Pi/Cursor) and upstream LLM providers (Gemini, DeepSeek, Zhipu z.ai).

Cryptographically signed policies that constrain AI agent behavior. Same attestation primitives as cilock, applied to agent execution: identity binding via MCP, JWT + signing-key separation, sublayouts for sub-agents.

Privacy-first PII tokenization middleware for LLM pipelines — local detection, encrypted vault, zero cloud dependency.

Guardrails for AI coding agents — block catastrophic tool calls before they run. Semantic, agent-neutral, near-zero false positives.

Session-layer circuit breaker for LLM & agent traffic — a transparent streaming proxy that meters every session and kills the one that goes rogue, without taking down your app.

Deploy AI-generated web apps to your own Kubernetes cluster in seconds, over MCP. No container build on the deploy path.

Per-operation approval and audit logging for secret access and git commit signing on Linux

A syscall-level security supervisor for AI agents.

Governed AI pipelines for service businesses. Deterministic-first, single Go binary, operator-approved.

Test, red-team, and deploy LLM applications with confidence. Multi-provider support (OpenAI, Anthropic, Gemini), MCP integration, self-play testing, and production SDK.

Guardrails for LLMs: detect and block hallucinated tool calls to improve safety and reliability.

Trust and verification for the skills your AI agent installs.

Local credential proxy: hand AI agents fake DSNs/API keys and swap in the real secret on egress, so credentials never enter agent context, logs, or traces.

CI release gate for AI agents. Blocks the pull request when a prompt, skill, MCP permission, or model quietly changes what your agent can do.

Tamper-evident audit log for Go: an append-only, hash-chained, optionally Ed25519-signed log/slog handler, with offline logx verify. Built on the standard library, zero third-party dependencies. Also a full slog toolkit (colored output, JSON, rotation, fan-out, CLI).

Sync gitignore-based rules to Claude Code sandbox, blocking file access via **all** tools (Read, Bash, Grep, etc.).

That install command your AI agent just ran? Never touched your host. tuprwre intercepts it, runs it in a throwaway Docker container, and shims the binaries back. Your agent gets its tools. Your system stays clean.