Today we’re launching one of our biggest updates yet: any agent you build in Pickaxe can now be deployed and sold as an MCP.
MCP lets AI tools connect directly to external agents and capabilities. With this release, your customers can add your Pickaxe to Claude Code, Codex, Cursor, VS Code, or any MCP-compatible client - and use it directly inside their existing workflow.
This unlocks two powerful new ways to use Pickaxe:
- Connect to the Pickaxe API through MCP. Bring your Pickaxe tools and capabilities into the AI coding clients you already use.
- Turn your Pickaxes into customer-facing MCPs. Build an agent, deploy it, and give customers a ready-to-copy configuration so they can install it in their own MCP client.
Getting connected only takes a few steps:
- Deploy your Pickaxe.
- Open the Connect an MCP client section.
- Choose Claude Code, Codex, Cursor, VS Code, or Generic HTTP.
- Copy the generated setup command.
- Paste it into your client and start using the Pickaxe as a tool.
This means you can now create specialized agents for research, data analysis, content creation, internal workflows, industry-specific tasks, or virtually anything else - then distribute and monetize those agents as MCPs through Pickaxe.
No custom MCP server setup. No complicated configuration for your customers. Just build, deploy, copy, and connect.
Pickaxe is now the easiest way to build and sell MCPs.
Try it with your next deployment and let us know what you build! 
4 Likes
It’s a huge feature, congrats for the job !
This feature is compatible if my pickaxe is an form or form+Talk ?
I imagine that the input field do not appear inside the harness provider
this is super exciting! my audience is not very technical (nor am I) and I’m struggling a bit on how I can explain this to my audience and how I can help them set this up once they purchase? Is there more guides or documentation on this? Thanks!
Hi @brandoo ,
Thank you for your question!
Chat and Form Pickaxes work via MCP. With a Form, the original visual form will not appear inside the MCP client. Instead, the client receives its fields as structured inputs.
Form+Chat is partially supported today: the form and initial response work, while follow-up chat is not yet available. Our engineers are working to extend MCP support so Form+Chat conversations can continue after the initial response.
1 Like
Hi @userresearchstrategi,
Thank you for your question!
A simple way to explain it is: MCP lets your customers use your Agent as a tool inside an AI application they already use, such as Claude Code, Codex, Cursor, or VS Code.
To help a customer connect, open your Agent’s API/MCP deployment and expand Connect an MCP client. Choose their client, copy the generated setup, and share it with them securely. They can then paste it into their client to add your Agent.
Our current API/MCP guide covers the available connection options. We do not yet have a dedicated setup guide written specifically for nontechnical customers, but we appreciate the feedback and agree that one would be valuable.
If your audience does not already use an MCP-compatible client, a Portal or Direct Link may be a simpler way for them to access your Agent.
If you would like personalized guidance on selling your Agents or managing end-user access more broadly, you can email us at info@pickaxe.co or schedule a one-on-one call.
