Esta página se muestra en inglés. Una traducción al español está en curso.
SkillsMay 14, 2026·2 min de lectura

ZenGuard — Runtime Guardrails for AI Agents

A real-time trust layer for agents with prompt-injection/PII/secrets detectors and tiered access; verified 150★, pushed 2026-02-03.

Agent Toolkit
Agent Toolkit · Community
Listo para agents

Staging seguro para este activo

Este activo primero queda en staging. El prompt copiado pide inspeccionar los archivos staged antes de activar scripts, config MCP o config global.

Stage only · 29/100Política: staging
Superficie agent
Cualquier agent MCP/CLI
Tipo
Skill
Instalación
Stage only
Confianza
Confianza: Established
Entrada
Asset
Comando de staging seguro
npx -y tokrepo@latest install 0f14bdd7-e715-5b6a-846b-b555960c79dc --target codex

Primero deja archivos en staging; la activación requiere revisar el README y el plan staged.

Introducción

A real-time trust layer for agents with prompt-injection/PII/secrets detectors and tiered access; verified 150★, pushed 2026-02-03.

Best for: Teams adding runtime safety checks (injection, PII, secrets) around tool-using agents

Works with: Python apps and agent frameworks; README links LangChain + LlamaIndex integrations and docs.zenguard.ai

Setup time: 10-25 minutes

Key facts (verified)

  • GitHub: 150 stars · 21 forks · pushed 2026-02-03.
  • License: MIT · owner avatar + repo URL verified via GitHub API.
  • README-backed entrypoint: pip install zenguard.

Main

  • Start with one detector policy per risk (prompt injection vs secrets vs PII) so you can tune false positives separately.

  • Pick the tier intentionally: README describes BASE vs DEDICATED tiers; use the free tier to validate integration first.

  • Wrap the agent at runtime: put checks around tool calls and retrieval inputs so the agent sees sanitized context.

  • Log decisions and outcomes: guardrails are most useful when you can audit why something was blocked or redacted.

Source-backed notes

  • README lists detectors such as prompt injection, PII, banned/allowed topics, keywords, and secrets (with Colab links).
  • README shows install commands pip install zenguard and poetry add zenguard.
  • README states that as of 03.07.2025 multi-detector detection is supported only on the Dedicated tier.

FAQ

  • Do I need a ZenGuard API key?: README examples reference ZEN_API_KEY for credentials.
  • What tiers exist?: README describes BASE (trial/rate-limited) and DEDICATED (higher QPS; standard/enterprise).
  • Can I use multiple detectors together?: README says multi-detector detection is Dedicated-only as of 2025-07-03.
🙏

Fuente y agradecimientos

Source: https://github.com/ZenGuard-AI/fast-llm-security-guardrails > License: MIT > GitHub stars: 150 · forks: 21

Discusión

Inicia sesión para unirte a la discusión.
Aún no hay comentarios. Sé el primero en compartir tus ideas.

Activos relacionados

Agent Governance Toolkit — Policy Guardrails for Agents

Microsoft's Agent Governance Toolkit adds policy checks, red-team scans, evidence verification, and runtime guardrails to autonomous agents.

Skills
Agent Toolkit

IronCurtain — Secure Runtime for AI Agents

IronCurtain is a runtime boundary for agents: it treats the model as untrusted and enforces policy for tool calls, writes, and network effects.

Skills
Script Depot

Pydantic AI Shields — Guardrails for Pydantic AI

Drop-in guardrail capabilities for Pydantic AI agents: cost budgets, tool permissions, and input/output guards; verified 63★, pushed 2026-05-05.

Skills
Agent Toolkit

Swarmclaw — Self-Hosted Multi-Agent Runtime + MCP

Self-hosted multi-agent runtime with memory, delegation, schedules, and MCP tools. Run swarms via web dashboard or CLI. Verified 481★; pushed 2026-05-14.

Skills
Agent Toolkit
◈Inicio🔍Buscar👤Yo