Integrations

Built to work where your agents live. Fencio plugs directly into your agent framework, vector store, and deployment pipeline. No rearchitecting required.

Platform integrations

  • OpenAI: Policy enforcement for OpenAI Assistants and function-calling agents.
  • Anthropic: Guardrails for Claude-powered agents with full tool-use inspection.
  • Google Gemini: Policy enforcement for Gemini-powered agents and function-calling workflows.
  • GitHub Copilot: Enforcement on Copilot's inline suggestions and agentic edits, blocking unauthorized repo writes and credential exposure before they land.
  • Cursor: Policy checks on every Cursor agent action, from file edits to terminal commands, before execution.
  • Claude Code: Native enforcement for Claude Code's file, shell, and repo actions, with full audit trails on every command.
  • Windsurf: Guardrails for Windsurf's Cascade agent, controlling file writes, terminal access, and dependency installs.
  • Microsoft Copilot: Enforcement on Copilot actions across Microsoft 365, blocking unauthorized document, mail, and data access before it happens.
  • LangGraph: Node-level policy enforcement across graph edges with full state observability.
  • LangChain: Drop-in middleware for LangChain pipelines, inspecting chains, tools and memory.
  • CrewAI: Multi-agent crew orchestration with per-role trust boundaries and audit logs.
  • OpenClaw: Community agent framework with extensible hook points for policy enforcement.
  • n8n: Workflow automation with policy checks on every agent-driven node execution.
  • MCP: Wrap any MCP server with enforcement on every tool invocation and resource read.
  • LangFuse: Bi-directional integration: push Fencio decisions into your Langfuse traces.
  • LlamaIndex: Policy-aware retrieval and query pipelines for LlamaIndex agents.
  • Haystack: Pipeline-level enforcement for Haystack agents and RAG workflows.
  • Plain HTTP/HTTPs Agents: Connect any HTTP-based agent via standard REST endpoints with zero SDK overhead.
  • Claude Plugins: First-class support for Anthropic's Claude plugins with full tool-use inspection.
  • Python Agents: Lightweight Python SDK for wrapping any agent logic with Fencio enforcement.
  • ChromaDB: Monitor vector store reads and prevent unauthorized RAG data exfiltration.
  • Pinecone: Enforce query-level policies on Pinecone indexes with under 30ms overhead.
  • Claude Cowork (coming soon): Native enforcement layer for Claude Cowork agents, policy-governed collaboration with full audit trails.

Deployment integrations

  • Gateway / Proxy: Inline gateway on the agent execution path. Every call inspected before it leaves your network.
  • API: REST API for policy evaluation from any runtime, in any language.
  • SDK: Embed enforcement directly in your agent code with a few lines of integration.
  • Runtime Middleware: Drop-in middleware intercepting tool calls and state transitions at runtime.
  • Sidecar: Per-pod sidecar enforcement alongside your agent containers, zero code changes.
  • Observability Integration: Stream enforcement decisions and audit trails into your observability stack.
  • Docker: Pre-built container images for sidecar and gateway deployment patterns.
  • Kubernetes: Helm chart with auto-scaling, readiness probes, and namespace-scoped policies.