About CapaKit#

When you let a coding agent write code, pull dependencies, and run scripts, you are delegating trust while keeping the responsibility. You shouldn't have to choose between moving fast with agents and keeping basic control over your host machine.

Why CapaKit exists#

Most of us inspect the final result and treat the app as a black box. Most security tooling does the same: it sandboxes the app at runtime and ignores the build phase. But in an agent-driven workflow, the build phase is where the supply-chain risk lives, install scripts running with full host access, secrets baked into config, dependencies pulled without review.

CapaKit puts a boundary around the whole lifecycle, build to run. No inherited host environment. No network access unless explicitly allowed. Ephemeral, single-use sandboxes for every build and run. Secrets resolved on demand instead of hardcoded.

About Me#

Roman Landenband I'm Roman Landenband, a software engineer in Tel Aviv. Over the last 20 years I've built backends, mobile apps, developer tools, and SaaS platforms.

I've been building with LLMs since GPT-3. When Anthropic announced MCP, I started hacking on what became CapaKit (originally mcpgate.com) to solve the security and integration friction I was hitting myself. I've been working on it full time since early 2025, without outside funding.

Read the Security Model

Back to top