In late January, the Algorand DevRel team released algorand-agent-skills – Agent Skills and MCP configs that help AI coding agents write correct Algorand code. While working on this, I've been thinking about what a purpose-built agentic development stack would look like, something designed specifically for the build-test-debug-deploy loop on Algorand, tightly integrated with the skills, and secure by default.
So I built VibeKit.
VibeKit is a CLI that sets up your AI coding agent for Algorand development with one command. It installs Agent Skills for knowledge, MCP servers for documentation lookups, and MCP tools that let your AI interact directly with the blockchain. Together, these layers let the agent handle real development tasks end-to-end, from prompt to mainnet.
What VibeKit sets up
Run vibekit init, and it detects your AI tools (Claude Code, OpenCode, Cursor, VS Code) and configures:
Agent Skills – Several structured markdown files covering smart contract development, testing, deployment, frontend integration, and more. Same skills from the algorand-agent-skills repo, now installed automatically.
Documentation MCPs – Your choice of Kappa (official Algorand docs) or Context7. When the skills don't cover something, your agent can look it up.
Development MCPs – Blockchain operations across five categories:
-
Contracts: deploy, call, introspect ABI
-
Assets: create, transfer, freeze
-
Accounts: fund, switch, send
-
State: global, local, boxes
-
Indexer: search transactions, query logs
Now your AI can build a contract in TypeScript or Python, deploy it to LocalNet or TestNet, create test assets, call methods, and verify state – all in one conversation.
Keys never touch the AI
This was non-negotiable. I built wallet providers into VibeKit, so your private keys never get exposed to the LLM.
You can use an experimental HashiCorp Vault for key management, or your OS keyring for local development. The AI requests transactions, and the wallet provider signs them. The AI never sees the keys.
The agentic stack
VibeKit completes the development stack. Skills provide knowledge. Documentation MCPs enable research. Development MCPs enable action. And wallet providers keep it secure.
Your AI can now own the full loop: write, compile, test, deploy, debug, iterate.
Getting started
Install VibeKit:
Run the setup wizard:
Check that everything's working:
Open your AI tool and start building.
Website: https://getvibekit.ai
GitHub: https://github.com/gabrielkuettel/vibekit
MIT licensed. This is an early release – feedback and contributions welcome.
Trust, but verify
I should add that none of this means Algorand smart contracts can be "vibe coded" without being reviewed. Contracts that hold financial value need to follow best practices and require real software engineering experience. VibeKit accelerates your workflow; it doesn't replace your judgment.
On a personal note: while I do believe the era of manually writing code is winding down, I've noticed that even frontier models need to be watched like a hawk. Claude Code with Opus 4.5, for example, has a sneaky tendency to let bad architecture decisions pile up. The code almost always works, which makes it tempting to overlook the accumulating technical debt until it's a problem.
That doesn’t mean you can’t use these tools to learn or build a working prototype, even with limited technical experience. In fact, I would encourage you to do so! Just keep in mind: you're still responsible for what gets deployed.
Disclaimer:
The content provided in this blog is for informational purposes only. The information is provided by the Algorand Foundation and while we strive to keep the information up-to-date and correct, we make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability with respect to the blog or the information, products, services, or related graphics contained in the blog for any purpose. The content of this blog is not intended to be legal, financial, or investment advice nor is it an endorsement, guarantee, or investment recommendation. You should not take any action before conducting your own research or consulting with a qualified professional. Any reliance you place on such information is therefore strictly at your own risk. All companies are independent entities solely responsible for their operations, marketing, and compliance with applicable laws and regulations. In no event will Algorand Foundation nor any affiliates be liable for any loss or damage including without limitation, indirect, or consequential loss or damage, or any loss or damage whatsoever arising from loss of data or profits arising out of, or in connection with, the use of this blog. Through this blog, you may be able to link to other websites which are not under the control of the Algorand Foundation. We have no control over the nature, content, and availability of those sites. The inclusion of any links does not imply a recommendation nor endorse the views expressed therein.