Switch to Symbiotic in minutes
Coming from Claude Code, Copilot, or Codex? Install, sign in, and pick up where you left off. Your instructions file, skills, and MCP servers are read as they are. First prompt in about 15 minutes.
Three steps. Nothing to rewrite.
Symbiotic Code reads the setup you already have. Install it, point it at your model, open a project. You're working in minutes.
Available as a terminal CLI or a desktop app. Grab whichever suits you and start using it straight away
Everything you had. Plus security.
Symbiotic Code is built on OpenCode, so the workflow is the one you know: prompt, plan, approve, review the diff. Underneath it, your team's security rules travel with the agent and risky commands run inside a sandbox.
Your muscle memory still works
Same as any coding agent you've ever used: you prompt, it plans, you approve, you review the diff. The security work happens automatically and seamlessly behind the scenes.
Your extensions come with you
Skills and MCP servers load from config without changes, and OpenCode plugins work as-is. All of it runs under the same permissions as the agent.
Your models, your keys
Start on the included model, or bring Claude, GPT, Gemini, or a local one. Routing can pick the right model per turn instead of locking you to a single choice.
Access the models you love, keep your costs down with turn-by-turn smart routing, and bring your own LLM for privacy
How it compares to Claude Code, Copilot, and Codex
You lose nothing you use today. You gain a harness that fixes vulnerabilities as the code is written, keeps secrets out of the prompt, and won't turn on you.
Questions & answers
Will this slow my developers down?
The opposite. In practice, security runs at generation time, so code arrives clean on the first pass: fewer review loops, no scan-fix-regenerate cycle, no rework tickets.
What's the difference between a coding harness and a coding agent?
An agent writes code. A harness runs the agent inside a loop you define: policies applied before generation, verification after, and a sandbox around every action. Symbiotic Code is the harness; the models are whichever you pick.
Does it work on code my team writes by hand?
Yes. The IDE plugin, PR apps, and CI integration scan and fix any code in the repo, not just what the agent produced.
What happens to my CLAUDE.md, skills, and MCP servers?
They carry over as-is. Most developers are productive within a few minutes.
GENERAL
Will this slow my developers down?
The opposite. In practice, security runs at generation time, so code arrives clean on the first pass: fewer review loops, no scan-fix-regenerate cycle, no rework tickets.
What's the difference between a coding harness and a coding agent?
An agent writes code. A harness runs the agent inside a loop you define: policies applied before generation, verification after, and a sandbox around every action. Symbiotic Code is the harness; the models are whichever you pick.
Does it work on code my team writes by hand?
Yes. The IDE plugin, PR apps, and CI integration scan and fix any code in the repo, not just what the agent produced.
What happens to my CLAUDE.md, skills, and MCP servers?
They carry over as-is. Most developers are productive within a few minutes.