guardrails

A Python runtime for decision systems where models propose and deterministic checks decide, with replayable traces, hard checks, and support for agents, knowledge, and guarded tool calls.

Open-source runtime guardrails for AI agents with cost limits, loop detection, policy enforcement, observability, and dashboard/CLI support.

Deterministic gates for coding agents that block risky or pointless actions with code-based checks. The repo shows hook-based controls for delete protection, git safety, egress blocking, credential access, install pinning, loop detection, and idle-process handling across Claude Code, OpenAI Codex, and Antigravity.

A Python library for agent evals and guardrails that uses typed decision questions instead of an LLM judge. It supports one-request evaluation for traces, with examples for tool choice, grounding, relevancy, and injection checks.

Browser-local tool that checks AI agent instructions for missing written guardrails against 23 public controls. It supports pasted prompts, local files, and public GitHub files, with CLI and GitHub Action options for recurring checks.

RiskKernel is an open-source, self-hosted reliability runtime for AI agents. It adds hard budgets, kill switches, crash-resumable runs, human approval gates, and OpenTelemetry export, with support for OpenAI, Anthropic, and existing agent stacks.

Open-source control plane for coding agents that turns GitHub issues into bounded runs with validation, review, rework, and human merge gates.

Open-source Rust LLM gateway for AI agents with full-fidelity traces, a tamper-evident audit ledger, offline verification, and in-request guardrails.

Local guardrails for AI coding agents that intercept destructive actions before they execute, now extended to Git hooks for commit and push enforcement.

Klent is an AI agent guardrails product that sits between agents and their tools, letting teams allow, deny, pause, rewrite, or redirect tool calls with policies, human approval flows, and audit trails.

An open-source governance layer and policy-decision daemon for autonomous AI agents, designed to enforce guardrails, budget limits, and runtime policy checks.

Go packages for building agents in Go, with a streaming runner, tool dispatch, guardrails, compaction, provider adapters, and verified completion helpers.

Local-first governance layer for AI systems that monitors prompts and responses to block secrets, PII, and runaway LLM spending before requests leave the machine.

A Blue41 case study on how a banking AI assistant could be abused through indirect prompt injection, and what mitigation layers help reduce the risk.

A Python filter layer for LLM apps that evaluates responses against a Lean-verified Scherf logic backend and can pass, flag, correct, or block output.

A comparison guide covering prompt injection detection and defense tools for enterprise AI agents, including production screening, red-team testing, guardrails, and action-boundary controls.

Giskard presents an AI red-teaming and continuous evaluation platform focused on finding hallucinations, security issues, and agent vulnerabilities. The page explains how it compares tools for agent-level testing, regression checks, and guardrail workflows.