YI GUARD

Check every agent action before it runs.

Yi Guard checks each tool call, MCP call and outbound request against your policy before it runs, inside your own process, and keeps a redacted record of every decision.

DEMO

See it work

Yi Guard · support agentExample

The agent is serving customer C-1001.

  1. Agent wants to look up order 1042

    Checking
    lookup_order(customer_id="C-1001", order="1042")

    It is this customer’s own order.

  2. Agent wants to look up order 2077

    Checking
    lookup_order(customer_id="C-2077", order="2077")

    It is someone else’s order, and no rule allows it.

  3. Agent wants to export every customer record

    Checking
    export_customers(fields=["name", "phone", "address"])

    It gets the masked export, with no phones or addresses.

  4. Agent wants to refund ¥199 on order 1042

    Checking
    refund(order="1042", amount=199)

    The agent stops and waits for you.

Rules

  • Orders: only this customer’s own
  • Customer exports: masked only
  • Refunds: a person approves
  • Everything else: denied

Decided locally. Arguments are never uploaded.

WHAT IT DOES

What Yi Guard does

Decided before the side effect

The SDK evaluates the policy locally, right before the tool runs. Raw arguments, results and model text stay in your process.

Five decisions

Allow, deny, allow with warning, require approval or degrade. When Guard can’t decide, the tool does not run.

Your framework

Adapters for LangChain, LangGraph, Claude Agent SDK, OpenAI Agents, Pydantic AI, Google ADK and more, plus the Vercel AI SDK in TypeScript.

MCP proxy

Put Guard between the agent and its MCP servers. Tools are admitted one by one, their metadata is pinned, and every call is checked.

Observe first, then enforce

Start in Observe mode, see what the agent really does, generate a least-privilege policy and preview it before anyone publishes.

Evidence you can export

Each decision carries the policy version, rule and reason. Pull events and the audit log as NDJSON into your own log system.

HOW TO START

How to start

  1. 01

    Connect an agent

    Wrap its tools with the SDK, or put the MCP proxy in front of its MCP servers.

  2. 02

    Observe

    Run in Observe mode and see what the agent really does.

  3. 03

    Publish a policy

    Generate a least-privilege policy from what the agent did, have a second person review it, then enforce it.

CODE → EVAL → GUARD · CLOUD

Other products

01 · BUILD

Yi Code

Tell it what to change. It edits the code in an isolated sandbox, runs the tests and opens a draft PR for you to review.

Explore Yi Code

02 · BEFORE LAUNCH

Yi Eval

Runs your agent for real in a sandbox, through normal tasks, tool faults and attacks. A version that fails the release gate does not ship.

Explore Yi Eval

MODELS · COMPUTE

Yi Cloud

Works with the OpenAI and Anthropic SDKs. Change the base URL and call models. Each sandbox and VM is a separate virtual machine, created and deleted through the API.

Explore Yi Cloud

All products

Next step

Want to try it on your own chips?

Book a PoC