Codezero Launches Cordon: One Command to Keep Credentials Safe Across AI Coding Agents

Cordon is the credential containment layer the industry never built – available first for Claude Code, Codex, and Hermes, with zero code changes required.

BEND, Ore. and VANCOUVER, BC, April 29, 2026 /PRNewswire/ -- Codezero today announced the launch of Cordon, a free, one-command security layer that protects developer credentials across every major AI coding agent.

The Gap Nobody Closed
The uncomfortable truth in security is that the industry has built incredible tools. Vaults, identity providers, policy engines, secret scanners. Our team has watched organizations do everything right with every one of them and still get breached.

Because as an industry we've been building more flashlights instead of solving the core plumbing problems by design.

AI coding agents are now indispensable. Claude Code, Codex, and Hermes write code, call APIs, run commands, and automate workflows at extraordinary speed. But to do their jobs, they need credentials and today's standard is to hand those over in plaintext: environment variables, .env files, shell history, MCP server configs. The credential enters the runtime. And once it's there, it's a target.

AI isn't just accelerating development. It's exposing us all to the weaknesses that exist across our software supply chains. A single agent workflow can chain dozens of API calls, each one touching a credential the agent has full access to. Those credentials live in runtimes, in plaintext, where they can be logged, leaked through prompt injection, exfiltrated via tool output, or copied across plugin boundaries. The attack surface isn't growing linearly, it's compounding with every model, every tool call, every autonomous system deployed.

The consequences are already materializing:

• A single audit of one major agent ecosystem uncovered 512 vulnerabilities (Argus Security Platform)
• This week: Bitwarden CLI npm package compromised to steal developer credentials
• Last week: Vercel Shinyhunters platform breach

And the industry is still responding with scanners that find leaks after they happen. You can't fix plumbing with cameras.

We can't keep shipping software the same way we always have and expect different outcomes in an agentic world, so we're releasing Cordon first to an initial cohort of users to make sure we solve the problem we set out to solve, before adding to an attack surface that's already growing faster than the industry can track.

What Cordon Does
Cordon sits at the network layer and when a credential is needed, Cordon pulls it from their existing vault – 1Password and macOS Keychain to start – injects it directly into the request in transit, then zeroizes it from memory.

The credential is never written to memory. Never loaded into the agent's environment. Never visible in a log or the model's context window. Agents can still do everything they're built to do. What they cannot do is see, access, or expose an organization's credentials.

Setup takes one command: cordon setup claude-code

No code changes. No new vault required. No ripping out the tools already in the stack because access and exposure aren't the same thing.

"A credential containment layer is essential to protect credentials for all enterprises that build software with or without an LLM. This approach significantly shrinks the attack surface for threat actors to use their favorite exploit, compromised credentials," said Jim Routh, former CSO, cybersecurity advisor, faculty member and mentor.

Who This Is For

• Security teams who've done everything "right" and still can't keep credentials out of runtimes.
• Engineering leaders scaling AI agents who've realized their attack surface just multiplied by 10x.
• Developers tired of being blamed for a structural gap no one built a solution for.
• Enterprises who need credential governance that works with their existing stack, not against it.

Supported at Launch

• Claude Code
• Codex
• Hermes

Compatible with 1Password and macOS Keychain, out of the box.

"Developers shouldn't have to choose between using the best AI agents and keeping their secrets safe. One command, and the credential risk disappears – regardless of which agent you're running," said Reed Clayton, CEO and Co-Founder of Codezero.

Availability
Codezero's Cordon is available today to an initial cohort of users. If you'd like to be an early testing partner, visit http://codezero.io.

About Codezero
Codezero is building toward a world where no agent, developer, application, or pipeline ever sees a credential it uses where containment is so seamless, so universal, and so structurally embedded in how software runs that a credential breach becomes an artifact of a previous era.

Cordon's credential containment layer scales across every runtime, agent, and pipeline without replacing a single tool already in your stack. Its architecture is vault-agnostic, identity-provider-agnostic, and activates with a single command.

SOURCE Codezero