Skip to main content

imperal-mcp

PyPI Python

Python 3.7 – 3.14 · pip install imperal-mcp works on every one of them.

Build, ship, and run real cloud apps on Imperal — straight from your AI agent.

imperal-mcp puts the app-building power of Imperal Cloud — the ICNLI AI Cloud OS that Webbee 🐝 runs on — inside any MCP-capable agent. Connect it to Claude, Claude Code, Codex, Cursor, Gemini CLI, Cline, Windsurf, or Goose, sign in with your own Imperal account, and your agent can design, deploy, and operate real cloud apps for you — data, logic, and UI included — without ever leaving the chat.

You describe what you want. Your agent builds it. Imperal runs it.


What you can do

Tell your agent what you want — and it builds it on Imperal, end to end:

  • Design an app from a description — its data/entities, the functions that act on them, and the UI panels that render them, as one clean declarative spec.
  • Validate it instantly against the live Imperal schema, before anything ships.
  • Smoke-test a function in an isolated sandbox and see it actually work.
  • Deploy it live to your Imperal account — instantly available, fleet-wide.
  • List & inspect every app you've built.
  • Run your apps — query their data through their own tools, right from the agent.

Apps your agent can build in a single flow:

  • a habit / workout / reading tracker with list + detail views
  • a personal CRM or contacts book
  • a snippet vault or bookmark manager
  • a lightweight expense or invoice log
  • a notes / tasks / journal app
  • …any data-backed app with screens and actions you can describe in plain words.

How it works

Your client's own LLM authors the app; imperal-mcp gives it the rails. It reads the IR spec and worked examples, validates the draft against the live schema, smoke-tests it, and deploys it to the cloud — no hand-written deployment scripts, no YAML wrangling. The heavy lifting — orchestration, hosting, scaling, the kernel — is Imperal's. Your agent just describes the app; the kernel makes it real.


Quick-add to your agent

Client One-liner
Cursor Add to Cursor
Claude Code /plugin marketplace add imperalcloud/imperal-mcp then /plugin install imperal-mcp@imperal
Everything else See per-client config

Install imperal-mcp and sign in first — see below.


Install & sign in

pipx install imperal-mcp      # or: uv tool install imperal-mcp   or: pip install imperal-mcp
imperal-mcp login             # opens your browser to sign in to Imperal

Works with any Python from 3.7:

  • Python 3.10 or newer — imperal-mcp runs directly on it.
  • Python 3.7 – 3.9 (macOS ships 3.9, Ubuntu 20.04 has 3.8) — the install is light, and the first imperal-mcp command sets up the same release on its own Python 3.12, once (about 30–45 s, needs internet; it fetches uv to do it). Every start after that is immediate. It lives in ~/.local/share/imperal-mcp/runtime and changes nothing else on your system. Run imperal-mcp login once before adding it to your agent, so that setup is done before the agent first starts the server.
Install problems
You see Why Do
error: externally-managed-environment pip install into the system Python is blocked on Homebrew, Debian 12+, Ubuntu 23.04+ pipx install imperal-mcp or uv tool install imperal-mcp
imperal-mcp: command not found after install the tool folder is not on PATH pipx ensurepath (or uv tool update-shell), then open a new terminal
Could not set up imperal-mcp on Python 3.12 on the first start Python 3.7–3.9 and no internet (or a proxy blocks github.com / pypi.org) allow those hosts and run imperal-mcp login again, or install uv and uv tool install imperal-mcp
No matching distribution found for imperal-mcp Python 3.6 or older install with uv — it brings its own Python: curl -LsSf https://astral.sh/uv/install.sh | sh, then uv tool install imperal-mcp
your agent says the imperal server failed to start the agent starts imperal-mcp from its own PATH, which may lack the tool folder — or its first start timed out during the one-time setup run imperal-mcp login once in a terminal; put the full path from which imperal-mcp in the agent's MCP config as command

login stores credentials in ~/.imperal/credentials.json and auto-refreshes your access token — no manual token management. Sign out any time:

imperal-mcp logout

That's it — every supported client uses the same imperal-mcp binary and the same stored sign-in.


Your first app (60 seconds)

  1. Install + imperal-mcp login (above), and add it to your agent.

  2. Ask your agent, in plain language:

    "Build me a habit tracker — I want to add habits, mark them done each day, and see today's summary. Then deploy it."

  3. Your agent authors the app, validates it, smoke-tests a function, and deploys it to your Imperal account. It comes back with the live app and its tools — ready to use.

Want to iterate? Just keep talking: "add a weekly streak view", "now build me an expense tracker too."


Per-client config

All clients use the same installed binary — imperal-mcp — and the same credentials from imperal-mcp login.

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "imperal": { "command": "imperal-mcp" }
  }
}

Claude Code

/plugin marketplace add imperalcloud/imperal-mcp
/plugin install imperal-mcp@imperal

Or add directly to your project .mcp.json:

{
  "mcpServers": {
    "imperal": { "command": "imperal-mcp" }
  }
}

Cursor

Click the Add to Cursor button above, or add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "imperal": { "command": "imperal-mcp" }
  }
}

Codex (OpenAI)

Add to ~/.codex/config.toml:

[mcp_servers.imperal]
command = "imperal-mcp"

Gemini CLI

{
  "mcpServers": {
    "imperal": { "command": "imperal-mcp" }
  }
}

Cline

In VS Code, add to Cline's MCP server list:

{
  "imperal": { "command": "imperal-mcp", "disabled": false }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "imperal": { "command": "imperal-mcp" }
  }
}

Goose (Block)

Add a StandardIO extension in Goose settings:

{ "type": "stdio", "name": "imperal", "cmd": "imperal-mcp", "args": [] }

Tools

Tool What it does
validate_ir(app_ir) Validate an app.ir.json — envelope + every declarative step. Returns {valid, issues}.
smoke_ir(app_ir, function, args?) Run one function in an isolated store and report {ok, result, trace}.
deploy_ir(app_ir, app_id) Deploy an app into your account (creates the app record if needed) — live instantly.
list_apps() List your developer apps.
get_app(app_id) Get one app's manifest + tools.
run_read_tool(app_id, function, args?) Run one of a deployed app's read tools to fetch its data.

Resources

URI Description
imperal://ir-spec The IR specification — structure, fields, action vocabulary.
imperal://ui-catalog All ui.* components usable in panels and render steps.
imperal://examples A worked app.ir.json (link-saver) to start from.

Prompt

build_imperal_app — step-by-step guidance for an LLM to go from intent to a deployed app: read the spec, author the IR, validate, smoke-test, deploy.


Configuration

Variable Default Purpose
IMPERAL_API_URL https://auth.imperal.io Auth / API base URL
IMPERAL_PANEL_URL https://panel.imperal.io Panel base URL
IMPERAL_TOKEN (not set) Optional token for CI / headless use — set it to skip browser login

For CI or headless environments, set IMPERAL_TOKEN to a token from panel.imperal.io → Developer → Access tokens.


Privacy & security

You sign in with your own Imperal account (browser OAuth). Credentials live locally at ~/.imperal/credentials.json (0600) and refresh automatically — nothing is shared with third parties and there's no telemetry. Read responses are PII-masked before they reach the LLM. Full details: PRIVACY.md.



Development

git clone https://github.com/imperalcloud/imperal-mcp
cd imperal-mcp
pip install -e '.[dev]'
python -m pytest

License

GPL-3.0-only — see LICENSE. Releases up to 0.5.5 were published under Apache-2.0.


Built on Imperal Cloud — the AI Cloud OS. 🐝

Metadata

Release files for imperal-mcp 0.5.10

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for imperal-mcp 0.5.10
File Size Uploaded Provenance
imperal_mcp-0.5.10.tar.gz 65.7 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for imperal-mcp 0.5.10
File Interpreter ABI Platform
imperal_mcp-0.5.10-py3-none-any.whl Python 3 none any Details

Total release size: 108.4 kB

Release files / imperal_mcp-0.5.10.tar.gz

Download URL imperal_mcp-0.5.10.tar.gz
Size 65.7 kB
Tags Source
SHA-256 checksum
How to use checksums
fbf227fc8a62bb042371720aa10e03b4d58e65cdd6950741797f1739e36b675a
BLAKE2b-256 checksum
How to use checksums
ad39bacb6422f90016b76f6689461f27e7944322b0761eb642e93b4a0e1a27d1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.16

Release files / imperal_mcp-0.5.10-py3-none-any.whl

Download URL imperal_mcp-0.5.10-py3-none-any.whl
Size 42.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
6ae86efea41aa0c366762e43b921f18ab1508d707085d4a7d8ac2997ed811ecd
BLAKE2b-256 checksum
How to use checksums
f9e0308f0f83289a8493bdea01ee0a49a8c5d4520113faf3fbd6df157892cc42
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.16

Security for imperal-mcp 0.5.10

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page