Guardrail and policy tools for agent security
Guardrails, policy engines and permission layers that limit what agents are allowed to do. Of the 53 tools on our agent security and safety page, 21 fit this narrower job. By pricing: 14 open source, 1 source-available, 4 free or freemium, 2 paid, usage-based or enterprise. The most-starred open repos here are LiteLLM (61,465 stars), Portkey AI Gateway (13,214 stars), Guardrails AI (7,529 stars). Most come from the security & governance (8), MCP servers (4), agent infra (3) categories.
Updated
LiteLLM
Open-source AI gateway and Python SDK that calls 100+ LLM providers in the OpenAI format, with cost tracking, guardrails, load balancing and logging.
Added
Portkey AI Gateway
Fast open-source AI gateway that routes requests to hundreds of LLMs through one API with fallbacks, load balancing and built-in guardrails.
Added
Guardrails AI
Open-source framework for validating LLM inputs and outputs with configurable guardrails.
Added
Superagent
SDK for AI agent safety that blocks prompt injections and unsafe tool calls, redacts PII and secrets, and runs red-team scenarios.
Added
Agent Governance Toolkit
Microsoft toolkit for policy enforcement, identity and sandboxing for AI agents.
Added
Failproof AI
Observability and policy enforcement for AI agent harnesses.
Added
coucou
Mac notch and phone companion that monitors your Claude Code, Codex, Cursor and Gemini CLI sessions and lets you approve permissions on the go.
Added
Archestra
Enterprise AI platform with guardrails, an MCP registry, gateway and orchestrator.
Added
Mercury Agent
Self-hosted personal AI agent with a written personality, local memory, permission-gated tools and token budgets, reachable via CLI or Telegram.
Added
Metorial
Open-source identity and access layer that connects AI agents to 1,200+ integrations via MCP, CLI or API with auth, permissions and observability.
Added
ToolHive
Stacklok platform that runs MCP servers in isolated containers with per-request identity, access policy and observability.
Added
Agent Safehouse
macOS sandbox for local coding agents that limits file and integration access with deny-first policy profiles.
Added
OpenAPPA
OpenAPPA is a deterministic guardrail that sits between an AI agent and its tools and tracks how data flows across tool calls, instead of asking a second model to judge each call. Policies decide whether untrusted or sensitive data may reach a given action, which blocks data exfiltration caused by prompt injection or hallucination while letting agents finish their tasks. It plugs into Claude Code, Pi, OpenCode, kAgent and Archestra, comes with reusable policy "batteries", validation, observability and reporting, and is backed by a NeurIPS workshop paper with published benchmarks. It is in preview as an open RFC.
Added
Rogue
Agent evaluator and red-team tool that probes your agent with test conversations and attacks and reports pass or fail against your policies.
Added
HOL Guard
HOL Guard is a free, open-source (Apache-2.0), local-first security layer for AI coding agents such as Claude Code, Codex, Cursor and GitHub Copilot CLI. On supported integrations it checks risky shell commands, reads of secret-bearing files, package installs and new or changed MCP tools before they run, lets you allow, ask or block each action, and keeps a local record for review. It installs with pipx (hol-guard) and works offline; optional Guard Cloud adds synced approvals, team policy and dashboards.
Added
Halofy
Halofy is an access and governance layer for a team's AI agents, giving every agent organizational context through one identity and policy model with provenance and audit, alongside a desktop AI workstation that turns power users' agent practices into shared policies.
Added
Credal
Spins up governed MCP servers from your company systems so Claude, ChatGPT, Cursor or your apps can query them with permissions.
Added
NVIDIA Open Agent Safety Platform
Reference stack pairing OpenShell with NVIDIA Sentry on BlueField-4 DPUs for hardware-level agent monitoring.
Added
Openfort
Open-source stablecoin wallet infrastructure with policy-gated agent wallets, gas sponsorship and x402 support.
Added
Privy
Embedded and server wallets developers can hand to agents with policy controls.
Added
Zenity
Security and governance for AI agents across enterprise platforms.
Added
GitHub repos (16)
| Repo | GitHub stars | Licence | Last push |
|---|---|---|---|
| LiteLLM ↗ | 61,465 | MIT | 2026-10-21 |
| Portkey AI Gateway ↗ | 13,214 | MIT | 2026-05-25 |
| Guardrails AI ↗ | 7,529 | Apache License 2.0 | 2026-08-26 |
| Superagent ↗ | 6,770 | MIT | 2026-08-25 |
| Agent Governance Toolkit ↗ | 6,435 | MIT | 2026-10-21 |
| Failproof AI ↗ | 5,560 | MIT | 2026-10-20 |
| coucou ↗ | 5,468 | — | 2026-10-20 |
| Archestra ↗ | 4,373 | AGPL-3.0 by default, with enterprise-licensed files | 2026-10-21 |
| Mercury Agent ↗ | 3,433 | — | 2026-10-21 |
| Metorial ↗ | 3,367 | — | 2026-10-18 |
| ToolHive ↗ | 2,293 | — | 2026-10-21 |
| Agent Safehouse ↗ | 2,110 | — | 2026-10-19 |
| OpenAPPA ↗ | 1,596 | — | — |
| Rogue ↗ | 1,069 | — | 2026-05-04 |
| HOL Guard ↗ | 947 | — | 2026-10-21 |
Top 15 of 16 by stars. Stars and licence come from each repo, checked when the listing was last verified.
FAQ
What are the best AI tools for agent security and safety focused on guardrails and policies?
We list 21. The first three by GitHub stars and featured status are LiteLLM: Open-source AI gateway and Python SDK that calls 100+ LLM providers in the OpenAI format, with cost tracking, guardrails, load balancing and logging.; Portkey AI Gateway: Fast open-source AI gateway that routes requests to hundreds of LLMs through one API with fallbacks, load balancing and built-in guardrails.; Guardrails AI: Open-source framework for validating LLM inputs and outputs with configurable guardrails.
Are there free or open-source options?
Yes: 14 open source, 1 source-available, 4 free or freemium, 2 paid, usage-based or enterprise. Pricing is as each maker states it.
More from agent security and safety
Tools are grouped by what each says it does. Missing one? Submit it free.