AI guardrails & agent security
3,625 agents · most starred first
AI guardrails & agent security, page 28 of 37
NullBridge AI Agent Identity Governance SDK
Core runtime, kernel, and trust layer for the Agent Governance Toolkit
LLM Prompt Injection Detection API — 5-layer detection (Vigil + DistilBERT + mDeBERTa + Rules + Groq)
Authorization, delegation, provenance, and verifiable-audit engine for AI agents
Audit & Compliance SDK for AI Agents
Library-first authentication and authorization SDK for AI agents
Portable, stdlib-only guardrail for autonomous coding agents — safe continuations instead of reckless actions.
OpenClaw integration package for Radware Agentic AI Protection
Guard LLM tool calls with rules, scoring, and audit trails.
Agent Token Budget MCP — per-session spend cap with signed budget-exhausted attestations. Twin of bft-progress-council-mcp (spend-axis vs stall-axis guardrail). By MEOK AI Labs.
Approve or deny your AI coding agent's risky commands from your phone, with an audit trail. Works with Claude Code and OpenAI Codex.
Static preflight scanner for MCP servers and agent skills
Cross-agent permission policy specification for AI coding agents
NDK - Nostr Development Kit. Includes AI Guardrails to catch common mistakes during development.
Circuit breaker SDK for LLM agents.
Official Python SDK for AgenticDome security guardrails, DLP, tool authorization, and multi-agent delegation enforcement.
Eight-layer middleware guardrail pipeline for LLM-powered personal finance applications
Provider-agnostic persuasion guard built with LangChain and LangGraph.
AgenticDome security middleware and firewall plugin for OpenClaw.
Security scanner for AI agent skill files
Constitutional Governance Kernel for AI Agents — trust scoring, approvals, audit trail
This is a reserved placeholder package to mitigate dependency confusion risks.
Zero-dependency LLM hallucination detection middleware with billing & dashboard — real-time fact-checking proxy for LLMs
Alpha agent trust middleware: audit trails, safety guardrails, HITL, and tool validation
Policy kernel SDK for verifying autonomous AI agent actions before execution.
AgentAdmit SDK: user-mediated AI agent authorization for any app
A safety layer that lets developers use AI coding agents at work without triggering security alerts
FastAPI middleware for verifying agent requests using the Agentis identity platform
Drop-in singulr agent SDK — one import protects any LLM agent via Singulr Gateway
Runtime governance for autonomous AI agents — per-run tool allowlists, write ceilings, loop detection, step limits, and timeouts. Works with LangGraph, OpenAI Agents SDK, or any TypeScript agent framework.
Agent Manifest SDK — cryptographically anchor all 10 artifacts defining an AI agent at deployment
Deterministic governance middleware for LangChain agents — policy enforcement, cost limits, tool allowlisting, NHI scope controls, and SARIF audit evidence. No LLM in the governance path.
Policy-as-code guardrail enforcement for enterprise LLM applications
Python SDK and drop-in LLM guardrail for the SentraGuard backend (OpenAI / Anthropic / Gemini compatible).
Zero-dependency TypeScript security filter for LLM streams — redact secrets, enforce tool policy, byte and event modes.
Zero-trust mathematical execution compiler for AI agent swarms.
A pluggable protective harness for conversational AI agents — drop-in OpenAI-compatible proxy that defends against jailbreaks, prompt injection, data exfiltration, and denial-of-wallet.
TRACE v0.2: hardware-attested governance records for AI agents
A red-team benchmark for AI agents in financial workflows: measures regulatory-control bypass (sanctions evasion, threshold structuring, dual-approval defeat) and the efficacy of guardrails that try to stop it.
Deterministic governance component for Haystack pipelines — policy enforcement, PII detection, cost tracking, and structured audit evidence. No LLM in the governance path.
Protocol-agnostic classification, risk tiers, and exact-call policy. Consumable by MCP, A2A, LangGraph, and Vercel AI tool runtimes.
FastAPI middleware for AgentVisa AI agent verification
LedgerProof adapter for the Anthropic Python SDK. Side-channel transparency receipts for EU AI Act Article 50.
n8n community node for Radware Agentic AI Protection in-path enforcement
AgentTrust ID SDK — runtime authorization, opaque agent tokens, and Guardian checks for AI agents
Official Python SDK for ZeroTrust Agents Control Plane API
The only agent execution layer that enforces governance before consequences land. x402-native paid execution, financial control plane, immutable receipts, output verification, EIP-712 portable reputation credentials, escrow settlement, RAG semantic memory
A2A TrustGate CLI — Safety, compliance, and governance for AI agents. Screen every action before it executes.
IAM for AI agents — permission and access control framework
Cryptographic receipts for every AI agent action. Hash-chained, tamper-evident, regulator-ready.
The emergency brake for multi-agent systems. Stop runaway LangChain & CrewAI agents in real time, before 50 cents becomes $47,000.
Provider-agnostic hard dollar budgets + loop detection for AI agent runs. Stops the spend before it happens.
Action-time proof and delegation verification for MCP agents
LedgerProof adapter for the Cohere Python SDK. Side-channel transparency receipts for EU AI Act Article 50.
LedgerProof adapter for Haystack 2.x — side-channel cryptographic receipts for EU AI Act Article 50 transparency.
LedgerProof adapter for the Mistral Python SDK. Side-channel transparency receipts for EU AI Act Article 50.
CI security regression testing for RAG apps and AI agents
SentraGuard framework-neutral guardrail core for AI agents — LangChain/LangGraph, LiteLLM, and direct provider adapters (no OpenTelemetry dependency).
Provider-agnostic credential routing and identity management for AI agents
Input and output guardrails middleware for Vercel AI SDK.
Policy-gated, durable, audited execution for AI agent tool calls. Every action gets a five-verdict policy check, a crash-safe checkpoint, and a tamper-evident audit event — a library, not a server.
Drop-in preflight guard for AI agents — one call to the x402 Trust Layer (guard + security grade + attestation) before any paid x402_fetch. Works with any agent framework (LangChain, CrewAI, Coinbase AgentKit).
Framework instalavel para agentes de IA que reduz contexto, roteia risco, valida entregas e registra evidencias, traces e limites; gates de ferramentas funcionam quando runner, MCP, gateway, hook, CI ou host os chama.
Reference verifier for the Agent Action Capsule SCITT profile (draft-mih-scitt-agent-action-capsule), with the typed Capsule carriers and their sealing boundary Capsule.seal(), and the anchor client.
Verifier for Agent Action Capsules — a SCITT application profile for verifiable agent actions (name claim; implementation to follow).
Governance layer for autonomous AI agents — pre-flight checks, runtime monitoring, and post-session reporting.
Enforcement and audit layer for AI agents — argument-level tool firewall, trust scoring, tamper-evident audit log, and prompt-injection detection across the Anthropic & OpenAI SDKs, LangGraph, and MCP.
Harness for measuring LLM agent resistance to indirect prompt injection and comparing defense effectiveness.
Anthropic Claude SDK with OmegaEngine governance - AI safety and compliance
OmegaEngine governance integration for autogen-omega
OmegaEngine governance integration for cohere-omega
FiGuard pre-flight spend authorization for LangChain agents
Verifiable, scope-gated tool execution for AI agent frameworks (IDProva)
Agent Passport Standard (APS) — Python SDK
Find and redact secrets in AI coding agent histories (Claude Code, and more).
Lightweight LLM API security layer — deterministic, no AI, no DB.
Wrap a deny.sh vault entry as a LangChain v1 tool. The credential resolves inside the tool boundary; only a narrowed DTO reaches the model. Fail-closed leak sweep.
Wrap a deny.sh vault entry as an OpenAI Agents SDK tool. The credential resolves inside the tool boundary; only a narrowed DTO reaches the model. Fail-closed leak sweep.
Dome Platform LangChain adapter — govern LangChain tools with Dome authorization
OmegaEngine governance integration for dspy-omega
Google Gemini SDK with OmegaEngine governance
Kyvvu compliance integration for Claude Code — policy enforcement via hooks
EMILIA Protocol guard for LangChain tools: a named human approves the exact irreversible action (Face ID / passkey) before it executes — fail-closed, with offline-verifiable Trust Receipts.
LangChain integration with OmegaEngine governance - callbacks, tools, and safety chains
llama-index tools MINT Protocol integration
MINT Protocol plugin for Microsoft Semantic Kernel — universal work attestation for AI agents.
Tenuo governance for Claude Code — warrants, hooks, MCP proxy, and Cloud lifecycle
Vendor-neutral control plane for AI agents — governance, audit, and runtime enforcement.
Earned autonomy for the OpenAI Agents SDK — warrantd drives the approval switch.
AI security skills grounded in mid-2026 threat reality, not stale framework documentation. 51 skills, 11 catalogs (439 CVEs / 177 CWEs / 805 ATT&CK + ICS / 170 ATLAS / 468 D3FEND / 8888 RFCs), 35 jurisdictions, 10-class catalog gap detector + budget gate,
AI coding firewall for Codex and Claude Code
SDK for obtaining scoped tokens for third-party services on behalf of apps or users. Authenticates the calling Vercel project via [`@vercel/oidc`](https://www.npmjs.com/package/@vercel/oidc) and exchanges the OIDC token for a Vercel Connect-issued credent
Pre-execution budget enforcement for agentic AI workflows
IRIS governance for Anthropic Claude — Cedar policy on every API call
IRIS governance for Google Gemini via google-genai
IRIS governance for OpenAI — Cedar policy on every API call
llama-index Bastion Prompt Protection guardrail integration
Authorization for AI agents: verifiable, scoped, revocable capability tokens (mandates) with attenuable delegation, for MCP and A2A. The reference implementation of agent authority — zero-dependency TypeScript & Python.
A runtime and definition-time security guardrail framework for AI agents and developers.
Governance proxy for every AI tool — PII/PHI masking, prompt injection detection, anomaly alerts, semantic cache, token quotas, vendor routing, compliance reports