Neo — AI/ML Engineering
Delegate AI/ML engineering tasks to Neo from your agent—submit work, monitor runs, and pull artifacts back into your repo.
Overview
Neo MCP is a MCP server for the Build phase that lets your agent submit and track Neo AI/ML engineering tasks and retrieve outputs in your project workspace.
What is this MCP server?
- stdio MCP server installable via uvx neo-mcp (PyPI v0.5.5)
- Submit Neo tasks, track execution, and retrieve outputs into NEO_WORKSPACE_DIR
- Requires NEO_SECRET_KEY (sk-v1-) from heyneo.com with optional staging via NEO_ENVIRONMENT
- Workspace dir must be git/project root so tasks read and write the correct files
- Server version 0.5.5
- stdio transport via uvx neo-mcp on PyPI
- Three documented environment variables including required NEO_SECRET_KEY
Community signal: 2 GitHub stars.
What problem does it solve?
Indie builders with ML ideas stall because orchestrating training, eval, and artifact handoff from the IDE requires too many manual steps outside the agent loop.
Who is it for?
Solo builders using heyneo.com who want agent-native task submission for AI/ML work rooted in a real git project directory.
Skip if: Pure frontend or non-ML products where you have no Neo account and no need for remote ML job execution.
What do I get? / Deliverables
Once Neo MCP is configured, your agent can queue Neo tasks against your repo root and pull completed outputs back for the next implementation step.
- Agent tools to create and monitor Neo AI/ML tasks
- Retrieved task outputs written under your configured workspace directory
Recommended MCP Servers
Journey fit
Neo MCP sits in Build where you extend the product with models, pipelines, and agent-assisted ML work tied to your project tree. Agent-tooling is the shelf for MCP servers that orchestrate external AI workers against your codebase.
How it compares
Remote AI/ML job MCP for Neo, not a local npm scanner or Chrome automation server.
Common Questions / FAQ
Who is neo-mcp for?
Developers with a Neo API key who build AI/ML features and want Claude Code or Cursor to submit and monitor Neo tasks on their project.
When should I use neo-mcp?
Use it when prototyping or building model pipelines, experiments, or agent features that Neo should run while your agent stays in the repo.
How do I add neo-mcp to my agent?
Install with uvx neo-mcp, set NEO_SECRET_KEY and NEO_WORKSPACE_DIR to your project root in the MCP env block, add the stdio server entry, and restart the client.