The coding agent your security team will approve. Full power on the desktop, governed on every call.
Kryex Code plans the work, edits your code, runs commands, drives a browser, hands pieces to sub-agents and remembers your project — on the model you choose. Anything that touches a governed system is checked on the server before it runs.
Free models included — no API key needed.
Start a task
Describe what you want done — the agent reads, edits, and runs commands in your workspace, with approval on anything risky.
Every step the agent takes, visible in one workspace.
The conversation sits in the middle. Around it, a dock you arrange yourself: the browser, a real terminal, every file it touched, the plan, background jobs, sub-agents and a side chat.
You decide how far it goes on its own.
From asking before every action to running freely, pick the mode that fits the task. The mode only changes when the agent stops to ask you — what it's allowed to touch is still decided by your grants on the server.
How modes and grants fit togetherA real coding agent underneath the enforcement.
Not a policy layer with a chat window attached to it. Kryex Code plans, delegates, remembers, and runs work on its own — the same execution boundary just applies to every one of these, the same way it applies to a single tool call.
Built for the whole job, not just the code.
.kryex/KRYEX.md (or an existing AGENTS.md) every turn, and keeps them up to date when you ask.Start on free models. Bring your own whenever you like.
Free models are included, with no API key needed. Or add your own key for 20+ providers, point it at your company's endpoint, or run fully local with Ollama or LM Studio. Kryex never marks up your provider's bill.
Bring your own modelIt asks for access. It never takes it.
Databases, MCP tools, the browser and web research all go through your organization's grants. When the agent needs more, it asks — and waits.
- Requests access in the conversation. READ, WRITE or ADMIN on a database, a browser or an MCP tool, with the reason attached, straight to an admin.
- Waits on held calls. When a tool needs approval, the agent follows it live and continues the moment an admin approves.
- Never holds the keys. Credentials for governed systems stay on the server; the app only sends which tool and which arguments.
- Keeps a local record. What the agent did on your machine is written to a local audit trail, alongside the server's own log.
- See what admins control in Kryex Admin
Two different problems.The same real agent.
Not mockups — real product chrome, run to the actual end. A database incident, a governed knowledge base — whatever the task looks like, it plans, reads, writes, and checks in with the gate before it acts, every time.