Solution · For enterprise

Your legal team needs a contract and a say in where agent data lives. Pick where it runs, on a yearly contract.

Enterprise runs on a yearly contract, on our cloud, a dedicated instance or your own Cloudflare account. One Protect policy covers the org.

Your requirement

We want contract terms, not a card.

Where it runs

  • Our gateway
  • Our cloud

What you get

  • A yearly contract
  • Requests set in the contract
  • Every feature

Enterprise Cloud. Your own Anthropic or OpenAI account pays the model. We add no fee on tokens.

We want a gateway that is ours alone.

Where it runs

  • Your own gateway
  • In our Cloudflare account

What you get

  • Its own hostname
  • Its own secrets
  • Every feature

Enterprise Dedicated, on a yearly contract.

Our data must stay in our account.

Where it runs

  • The same gateway
  • Deployed in your Cloudflare account

You hold

  • The data
  • The keys
  • The logs
  • The Cloudflare bill

There is no on-premises build.

An engineer's agent tries to read .env.

Blocked · never reaches your shellRead environment files

Set once by the org. An agent cannot loosen it.

Blocked by your team’s policy: Read environment files. .env files hold secrets.
Every member's agents follow it, on every machine.

Protect checks each tool call before your client runs it.

A script with an API key tries to turn Protect off.

API key: turn Protect off
403
A change that loosens the policy needs a person
signed in to the console, and a reason.

An API key can only tighten the policy. Every change goes into a hash-chained history.

The auditor asks what the agents tried.

A decision log row holds

  • Time and decision
  • Rule, tool and program
  • Host, project, session
  • Policy version

It never holds

  • Command text
  • Flag values
  • Environment values
  • File contents

Export as CSV or JSON Lines. Keep rows for 1 to 365 days.

The problem

Every agent has its own rules. No one place to set them.

Rules on each laptop

Each client has its own rules, set on each machine.

No record for the auditor

When security asks what the agents tried, there is no single log to show.

Terms you must set

Legal and security want a contract, and a say in where the gateway runs and who holds the data.

Where it runs

Three ways to run Enterprise. All on a yearly contract.

Enterprise CloudEnterprise DedicatedIn your Cloudflare account
GatewayOur gatewayYour own gatewayThe same gateway, deployed in your account
Runs inOur cloudOur Cloudflare account, with its own hostname and secretsYour Cloudflare account
What sets it apartContract termsA gateway that is yours aloneYou hold the data, the keys, the logs and the Cloudflare bill
ProtectOne policy for the orgOne policy for the orgOne policy for the org

From Plans and limits and Pricing. Requests are set in the contract. Each member's memory stays their own.

Proof

Tested on production. Read the run.

126 of 126Protect API checks pass on production
403for an API key that tried to turn Protect off
40-entry historyof policy changes checked out after the run

The run used single accounts. It did not test the org-wide Team policy.

Throwaway accounts in production, no customer traffic. Read the run →

How it works

One hop for every request. One policy on that hop.

  1. 01Connect each machine

    Each engineer runs npx @context-mode/cli. It connects Claude Code and Codex and keeps a backup of the old config.

  2. 02Pick where it runs

    Our cloud, a dedicated gateway, or your own Cloudflare account. The contract sets your requests.

  3. 03Read the log

    Every block goes into a decision log with its rule, tool and host. Export it as CSV or JSON Lines.

Every engineer's agent sends its requests through Context Mode. Protect checks each tool call against the one org policy, writes each block to the decision log, and the request goes on to the model on your own Anthropic or OpenAI account. The gateway runs on our cloud, a dedicated instance or your own Cloudflare account. Your agents each engineer Context Mode where you pick Model your account 1 Protect: one org policy 2 Decision log of each block 3 Our cloud, dedicated or yours
Your own Anthropic or OpenAI account pays the model. We add no fee on tokens.
Not built yet

What we do not ship today. So you can plan.

No SSO or roles

No SSO, SCIM or team roles yet. One Protect policy per account.

No on-premises build

The gateway runs on Cloudflare: ours, a dedicated instance, or your own account.

Two clients enforced

Claude Code and Codex. Other agents that go through the gateway are recorded, not enforced.

Only what passes the gateway

Protect checks only requests that go through it. Codex cannot be pinned to it today. See Keep the gateway on.

Reads the call, not the program

A script file, a compiled program or a payload decoded at run time gets past Protect. Keep your agent's sandbox on.

No automatic expiry

Stored prompts have no automatic expiry yet, and the console has no delete or export button. Ask us and we delete your account's data.

Products used

What this uses.

Protect

Checks every tool call before your client runs it. One policy for the whole org.

Team and Enterprise

Team gives one policy and one bill. Enterprise adds a contract and a choice of where it runs.

Enterprise deployment options

The three Enterprise ways, and what pauses at a limit.

Related: Secure coding agents, For platform teams, For startups.

One policy for every agent. Talk to us about the contract.

Write to us for a yearly contract. To try it first, npx @context-mode/cli connects your agents. 1,000 requests free, no card.

Get started free

Your coding agent costs less and remembers your project.

Start free: 1,000 requests, every feature, no card. After you sign in, the console gives you one command to connect Claude Code or Codex.

Read the docs