Neo — AI/ML Engineering

Neo MCP — AI engineering, without leaving your editor

Stay in Claude Code, Cursor, or Codex and hand your AI/ML work to Neo — an autonomous AI engineer that plans, builds, runs, and evaluates entire workflows from a plain-English request. neo-mcp is the Model Context Protocol server that connects Neo to the AI coding tools you already use, so you never switch tabs: ask your agent to send a task to Neo, and a local daemon writes the resulting code, models, metrics, and reports straight into your repo — on your machine, nothing stored remotely.

Because Neo is purpose-optimized for AI engineering — not a general-purpose coding assistant — it goes deeper on ML, LLM, and data workflows than a general coding agent can.

📚 Docs: https://docs.heyneo.com/neo-mcp · 🔑 Get an API key: Neo dashboard

What MCP unlocks

• 🧩 Stay in your editor — drive Neo from Claude Code, Cursor, VS Code (Copilot), Windsurf, Zed, Continue, or Codex. No new app, no context switching.
• 🔬 More depth — autonomous planning, experiments, evaluation, and iteration tuned for AI/ML, beyond what a generic coding agent attempts.
• 💾 Local-first — every output file is written to your machine; nothing is stored remotely.

What you can build with Neo

• 🤖 Generative AI & LLMs — RAG & semantic search, agents & chatbots, fine-tuning (Llama, Qwen, Gemma), document analysis
• 🧠 ML & deep learning — PyTorch / TensorFlow / scikit-learn training, architecture search, evaluation
• 📊 Data science & analytics — EDA, feature engineering, forecasting, segmentation, A/B testing, reporting
• 👁️ Computer vision — image classification, object detection, OCR
• 🎤 Speech & audio — speech-to-text, text-to-speech, audio classification
• 🔌 Bring your own keys — GitHub, HuggingFace, Anthropic, OpenRouter, OpenAI, AWS S3, Weights & Biases, Kaggle — stored locally, injected as env vars

For data scientists, ML & LLM engineers, analysts, researchers, and PMs who want results, not boilerplate.

---

Install

pip install neo-mcp

Requires Python 3.11+ on Linux, macOS, or Windows through WSL. Native Windows cmd.exe/PowerShell execution is intentionally unsupported because Neo commands use POSIX shell and container-path semantics.

Tip: use pipx install neo-mcp to install in an isolated environment and avoid conflicts with your project's virtualenv.

---

Use Neo from your editor

Neo runs in every major MCP-enabled AI editor — set it up once below. Replace sk-v1-YOUR_KEY with your actual key.

---

Claude Code

claude mcp add --scope user neo \
  -e NEO_SECRET_KEY=sk-v1-YOUR_KEY \
  -- neo-mcp

Open a new Claude Code session after running this. Neo tools load at session start.

Scope options: --scope user (global, recommended) · --scope project (writes .mcp.json in current repo) · --scope local (this machine only)

Verify it registered:

claude mcp list

---

Cursor

~/.cursor/mcp.json:

{
  "mcpServers": {
    "neo": {
      "command": "neo-mcp",
      "env": {
        "NEO_SECRET_KEY": "sk-v1-YOUR_KEY"
      }
    }
  }
}

Restart Cursor after editing.

---

Windsurf

~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "neo": {
      "command": "neo-mcp",
      "env": {
        "NEO_SECRET_KEY": "sk-v1-YOUR_KEY"
      }
    }
  }
}

Restart Windsurf after editing.

---

VS Code (GitHub Copilot)

.vscode/mcp.json in your workspace root (requires VS Code 1.99+):

{
  "servers": {
    "neo": {
      "type": "stdio",
      "command": "neo-mcp",
      "env": {
        "NEO_SECRET_KEY": "sk-v1-YOUR_KEY"
      }
    }
  }
}

---

Zed

~/.config/zed/settings.json:

{
  "context_servers": {
    "neo": {
      "source": "custom",
      "command": {
        "path": "neo-mcp",
        "args": [],
        "env": {
          "NEO_SECRET_KEY": "sk-v1-YOUR_KEY"
        }
      }
    }
  }
}

---

Continue.dev

~/.continue/config.json:

{
  "mcpServers": [
    {
      "name": "neo",
      "transport": {
        "type": "stdio",
        "command": "neo-mcp",
        "env": {
          "NEO_SECRET_KEY": "sk-v1-YOUR_KEY"
        }
      }
    }
  ]
}

---

OpenAI Codex CLI

~/.codex/config.json:

{
  "mcpServers": {
    "neo": {
      "command": "neo-mcp",
      "env": {
        "NEO_SECRET_KEY": "sk-v1-YOUR_KEY"
      }
    }
  }
}

---

Tools

Tool	Description
neo_submit_task	Submit a locally destructive AI/ML task; one active owner per workspace.
neo_list_tasks	List running and recent tasks — reconnects pollers automatically.
neo_task_status	Compact live backend-reported status; telemetry is not independent proof.
neo_get_messages	Read full task output when COMPLETED. Capped at ~20 000 tokens.
neo_send_feedback	Reply when Neo asks a question (WAITING_FOR_FEEDBACK).
neo_pause_task	Pause a running task.
neo_resume_task	Resume a paused task.
neo_stop_task	Stop and clean up a task permanently.
neo_list_integrations	List stored third-party API keys (names only — never the value).
neo_add_integration	Register a GitHub PAT / HuggingFace token / Anthropic key / OpenRouter key so Neo tasks can use it as an env var.
neo_test_integration	Call the provider's API to confirm a stored key is still valid.
neo_remove_integration	Delete a stored key from this machine.
neo_get_execution_evidence	Read bounded, hash-chained local execution observations.
neo_verify_task	Verify local artifacts with explicit file assertions and acceptance commands.

Integration tools store credentials locally (file 0o600 under ~/.neo/integrations/, or OS keyring with NEO_INTEGRATIONS_BACKEND=keyring). Keys never leave your machine. See the full guide at docs/INTEGRATIONS.md.

---

Environment variables

Variable	Required	Description
NEO_SECRET_KEY	Yes	Your Neo API key (sk-v1-...) from the Neo dashboard
NEO_DEPLOYMENT_ID	No	Pin a specific deployment UUID (auto-generated and persisted by default)
NEO_WORKSPACE_DIR	No	Override working directory (useful in Docker)
NEO_READ_ONLY	No	true — expose only status, messages, task/evidence reads, integration-test/list, and BYOK-list tools; reject every mutation and verification command
NEO_INIT_CHAT_TIMEOUT_SECONDS	No	Submit timeout in seconds; defaults to 60 to match the production VS Code client
NEO_POLLER_READY_TIMEOUT_SECONDS	No	Maximum pre-submit wait for the local daemon to establish its backend poll connection; defaults to 15

---

Troubleshooting

Symptom	Fix
neo-mcp: command not found	Re-run pip install neo-mcp and verify your PATH with which neo-mcp
Tools don't appear after registering	Open a new session — MCP tools load at session start, not mid-session
Invalid API key (401)	Re-check your key in the Neo dashboard
Trial or quota ended (403)	Top up at the Neo dashboard
Task submitted but no files written	Daemon failed to start — run neo-mcp doctor to diagnose
Status stuck on RUNNING	Call neo_task_status to check; run neo-mcp status to inspect the daemon