TweekIT MCP server
Commit Nov 10, 2026Normalizes and converts over 400 file types into AI-compatible formats to prevent ingestion failures in AI workflows.
Also known as equilibrium-team/tweekit-mcp, mcp
- Stars
- 1
- Tools
- 14
- Downloads, 30 days
- —
- Latest version
- 1.5.0
- Last commit
- Nov 10, 2026
- License
- MIT
What is TweekIT MCP?
TweekIT normalizes and converts over 400 file types into AI-compatible formats to prevent ingestion failures in AI workflows. It exposes 14 tools, such as extension, inext, outfmt.
Summary generated from the project's README
About TweekIT MCP server
TweekIT MCP Server acts as a universal media translator for AI workflows, designed to prevent ingestion failures caused by unsupported file formats. It allows developers and AI agents to normalize and transform over 400 different file types into standardized, AI-compatible formats. This service is particularly useful for users of Claude, ChatGPT, or IDEs like Cursor and Continue who encounter errors when uploading complex files like PSD, DWG, or various camera RAW formats.
Users can deploy the server locally using the Python SDK via pip or uv, or connect to a hosted remote endpoint. The system requires an API key and secret for authentication, which should be managed through environment variables or HTTP headers to ensure security. The server provides specific tools such as doctype, convert, and convert_url to handle tasks ranging from simple format changes to complex transformations like cropping, resizing, and background removal.
Written from the project's README. Every use case is checked against a quote from it.
How to install TweekIT MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall TweekIT MCP in Claude Code
$ claude mcp add --transport http mcp https://mcp.tweekit.io/mcp/Install TweekIT MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"mcp": {
"url": "https://mcp.tweekit.io/mcp/"
}
}
}Install TweekIT MCP in VS Code
.vscode/mcp.json
{
"servers": {
"mcp": {
"type": "http",
"url": "https://mcp.tweekit.io/mcp/"
}
}
}Install TweekIT MCP in VS Code CLI
$ code --add-mcp '{"name":"mcp","type":"http","url":"https://mcp.tweekit.io/mcp/"}'Install TweekIT MCP in Codex
$ codex mcp add mcp --url https://mcp.tweekit.io/mcp/Install TweekIT MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.mcp] url = "https://mcp.tweekit.io/mcp/"
Install TweekIT MCP in Gemini CLI
$ gemini mcp add -s user --transport http mcp https://mcp.tweekit.io/mcp/Install TweekIT MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"mcp": {
"url": "https://mcp.tweekit.io/mcp/",
"type": "http"
}
}
}TweekIT MCP tools 14
Detected from the READMEextensioninextoutfmtblobnoRasterizewidthheightpagealphabgColorurlfetchHeaders
Show all 14 toolsShow fewer tools
querymax_results
Configuration
TWEEKIT_ALLOWED_DIRSTWEEKIT_ALLOW_LOCAL_FILESTWEEKIT_API_KEYTWEEKIT_MINER_NOAUTH
Automated checks
Ran Nov 13, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachableequilibrium-team/tweekit-mcp is public and reachablegithub.com/equilibrium-team/tweekit-mcp ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-10github.com/equilibrium-team/tweekit-mcp ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model14 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNo tool named like delete, drop or execPass
- Remote endpoint answers the MCP handshakeinitialize answered with protocol 2025-06-18 in 204 msPass
README
View on GitHubTweekIT MCP Server & Python SDK
Ingest, Preflight, and Convert Just About Any Filetype Into AI Workflows — Official Open-Source Package by AMPS dba Equilibrium
Current version: v1.8.3
🤖 Claude & AI Hub Users: Tired of "unsupported file type" errors? Jump to our Claude File Format Enabler – a browser-based tool that converts 450+ formats (DOC, XLS, PSD, DWG, Camera RAW, etc.) into AI-compatible files in seconds, or use our built-in Python
preflight_checkandTweekitClientSDK directly in your pipeline.
Overview
TweekIT MCP Server is the universal media translator for AI workflows, making any file ready for processing in seconds. Built on the robust content processing engine of Equilibrium's MediaRich Server, a technology developed since 2000 and trusted by massive portals and media companies worldwide, TweekIT brings decades of expertise in handling complex media, drawing on a pedigree that includes the renowned DeBabelizer.
With TweekIT, you can upload, preview, transform, and download over 400 supported file types using a consistent API or widget interface. It removes the pain of unpredictable input formats by automatically normalizing and preparing assets for the next step in your workflow.
What is TweekIT?
TweekIT is a universal media ingestion and transformation service designed to take any supported file and make it AI ready in seconds. Whether you are normalizing formats, resizing assets, or extracting a single page from a complex document, TweekIT provides a consistent, API first way to get exactly the output you need without worrying about the quirks of individual file types.
Why use it in an MCP context?
In a Model Context Protocol (MCP) workflow, TweekIT acts like a universal translator for media. It sits between your AI agent and the unpredictable real world files it encounters, ensuring every asset is transformed into a compatible, standardized format before processing. This is similar to how GitHub Actions streamlines software automation, but applied to media handling in AI pipelines.
With TweekIT in your MCP toolset, your agents can:
- Accept a wider range of inputs from users without failure
- Automatically apply transformations such as cropping, resizing, format conversion, and background changes
- Pass clean, ready to use assets to the next step in your AI workflow with no manual intervention fixing the constant customer file ingestion failures we are all to familiar with
🤖 Claude Users: See Solving Claude's File Type Limitations for a dedicated guide on using TweekIT to overcome Claude's built-in file format restrictions and our standalone web converter tool.
Example scenario
A user uploads a scanned PDF of an ID card as part of an onboarding process. Your MCP agent calls TweekIT to:
- Convert the PDF to a PNG image
- Crop to just the ID card
- Resize to a standard 300x300 pixel headshot
- Return the transformed image directly into the verification pipeline
The entire process happens in seconds, and your workflow never sees an incompatible file format.
Key Capabilities
- Supports 400+ file types for seamless ingestion and conversion, leveraging a core engine refined over two decades to handle dynamic imaging and video processing.
- Stateless and API first design for fast, scalable integrations
- Widget or REST API access to fit any application architecture
- Enterprise grade security with secure key handling and short lived asset storage
- Instant compatibility fixes that prevent AI pipeline failures from bad input formats immediately
Try it now
- AI File Format Enabler – Browser-based converter for Claude, OpenAI, and other AI tools (DOC, XLS, PSD, DWG → PDF/PNG)
- Live Demo – Upload and transform files in your browser
- Use Case Examples – See practical workflows and code samples on the website. The specific MCP manifest conversions are shown in section 7 below.
Table of Contents
- Overview
- Solving Claude's File Type Limitations
- Requirements
- Installation
- Quickstart
- Authentication
- Server and Hosting
- On-Premises & Localhost Processing via Miner Renderery
- Client Compatibility
- Rate Limits and Pricing
- Core Concepts
- MCP API Reference
- REST API Reference
- Use Cases
- Demo + Show Code
- Error Handling and Troubleshooting
- Advanced Topics
- Registry Readiness
- Resources
Requirements
- Python 3.10 or later
- Docker (optional, for containerized builds)
httpxlibrary for HTTP requestsFastMCPfor tool registration and server functionality
Installation
Option 1: Install from PyPI (recommended)
# Using pip
pip install tweekit
# Using uv
uv pip install tweekit
# Run directly as an MCP server with zero local setup via uvx
uvx tweekit --transport stdio
# Using pipx / uv tool (isolated CLI environment)
uv tool install tweekit
Run the MCP server locally (tweekit and tweekit-mcp CLIs are both installed):
tweekit-mcp --transport streamable-http --port 8080
The CLI wraps server.py and accepts the same flags (--transport, --host, --port).
Python SDK Quickstart (tweekit / tweekit_mcp)
Installing tweekit from PyPI ships the full Python SDK (import tweekit or import tweekit_mcp) for both MCP client workflows and direct REST API integrations, plus the AI Hub preflight & modality router (powering Maitrix AI Hub multimodal file ingestion):
import asyncio
from pathlib import Path
from tweekit import (
TweekitClient,
TweekitDirectClient,
preflight_check,
)
# 1. Instant local preflight & AI provider format recommendation (no network roundtrip)
check = preflight_check("blueprint.dwg", target_provider="claude")
print(check["supported"], check["category"], check["recommended_outfmt"])
# -> True, 'design_vector_cad', 'png'
# 2. Direct Async Python SDK (calls TweekIT API directly via httpx)
async def run_direct():
async with TweekitDirectClient() as api:
result = await api.convert_file(
Path("presentation.ppt"),
outfmt="pdf",
output_path=Path("presentation_ai_ready.pdf"),
)
print("Saved:", result)
# 3. Async MCP Client (connects to https://mcp.tweekit.io/mcp or local stdio server)
async def run_mcp():
async with TweekitClient() as client:
tools = await client.list_tools()
batch_results = await client.batch_convert([
{"url": "https://example.com/scan.tiff", "outfmt": "png"},
{"path": "legacy_sheet.xls", "outfmt": "pdf"},
])
print("Tools:", tools, "Batch:", len(batch_results))
asyncio.run(run_direct())
Option 2: Clone the repository
git clone https://github.com/equilibrium-team/tweekit-mcp.git
cd tweekit-mcp
uv sync # or: pip install -r requirements.txt
Set environment variables as needed:
- PORT – Streamable HTTP port (default: 8080).
- PLUGIN_PROXY_PORT – Port for the ChatGPT plugin proxy (default: 8080 when running standalone).
- MCP_SERVER_PORT – Internal port for the MCP server when using
start_services.sh(default: 8000).
Quickstart
Step 1: Sign Up
Create your free TweekIT account and get started with 10,000 API calls included at no cost.
Pricing and Signup here to generate your API credentials in the MANAGE ACCOUNT page and start using the service.
Step 2: Configure Credentials Securely (Environment or HTTP Headers Only)
TweekIT MCP authenticates using your API Key and Secret configured in MCP client headers (for hosted Streamable HTTP) or TWEEKIT_API_KEY / TWEEKIT_API_SECRET environment variables (for local stdio).
Security Rule: Never place
TWEEKIT_API_KEYorTWEEKIT_API_SECRETinside natural-language prompts or LLM tool-call arguments. Credentials are intentionally excluded from MCP tool schemas so they never appear in model context or conversation logs.
Step 3: Make Your First API Call
Once your MCP client is configured with ApiKey and ApiSecret HTTP headers (or TWEEKIT_API_KEY / TWEEKIT_API_SECRET environment variables), call convert with only document and geometry arguments:
{
"method": "tools/call",
"params": {
"name": "convert",
"arguments": {
"inext": "png",
"outfmt": "png",
"blob": "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",
"noRasterize": false,
"width": 30,
"height": 30,
"page": 1,
"bgColor": ""
}
},
"jsonrpc": "2.0",
"id": 4
}
Step 4: See It in Action
You can explore TweekIT visually before writing any code. The live demo lets you upload files, apply transformations, and download the results instantly.
- Open the Live Demo
- View Use Case Examples for real workflows and code samples
Authentication
TweekIT supports multiple authentication methods (REST API, APP ID method is ONLY for using). Set up your API Key and Secret so you can start quickly while maintaining security best practices.
For MCP Server Use: API Key + Secret
This is the most secure and flexible option. Use this for all production deployments.
Steps:
- Generate your API Key and Secret in your TweekIT account settings.
- Store them securely — never commit to source control.
- Use server-side code or a reverse proxy to inject credentials into requests.
Security Tips:
- Always use a reverse proxy or other server-side method to protect credentials.
- Never embed secrets in client-side code.
Next Steps:
Once authenticated, you can:
- Upload media as part of the request using Base-64
- Recieve output
Server and Hosting
Hosted ready MCP server
All tools and resources are available here, used for testing and free trials:
- Public MCP endpoint (HTTP transport): https://mcp.tweekit.io/mcp
The currently available resource name is 'version'. The currently available tool names are 'doctype', 'convert', and 'convert_url'. See below for parameters, or query the MCP server as it will return metadata instructing use of each of these.
Versioning & Build Numbers
The project tracks semantic versions in major.minor.build format (for example, 1.8.3). Use scripts/bump_version.py to keep every metadata file aligned:
# Increment just the build number (e.g., 1.8.1 -> 1.8.3)
python scripts/bump_version.py --bump-build
# Set an explicit version
python scripts/bump_version.py --set 1.8.3
The script updates VERSION, pyproject.toml, server.py, server.json, claude/manifest.json, README.md, and uv.lock so container builds, manifests, and docs stay synchronized.
On-Premises & Localhost Processing via Miner Renderery
For regulated enterprises, confidential local workspaces, air-gapped networks, and workstations running the New Miner Installation Initiative, TweekIT's Miner Renderery (CGNMiner / MediaRich CORE™ + TweekIT_API local gateway) can be installed directly on-premises or on localhost.
When paired with a local Miner Renderery (http://127.0.0.1:16377/api/image/ or http://127.0.0.1:5000/api/image/), tweekit-mcp processes all 450+ document, image, CAD, RAW, and Office formats 100% on localhost with zero third-party cloud data egress:
# Launch tweekit-mcp against the localhost Miner Renderery in zero-cloud-egress mode:
TWEEKIT_ALLOW_LOCAL_FILES=1 \
TWEEKIT_ALLOWED_DIRS=/path/to/workspace \
TWEEKIT_MINER_NOAUTH=1 \
tweekit-mcp --transport stdio --local-miner
Key environment variables for On-Premises / Localhost Miner Renderery mode:
TWEEKIT_API_BASE_URL: Override the TweekIT API endpoint (defaults tohttp://127.0.0.1:16377/api/image/when--local-minerorTWEEKIT_MINER_LOCAL=1is set; container Kestrel profile useshttp://127.0.0.1:5000/api/image/).TWEEKIT_LOCAL_ONLY=1: Enforces zero cloud egress — verifies thatTWEEKIT_API_BASE_URLresolves strictly tolocalhost(127.0.0.1,::1,localhost) or an allowlisted internal node inTWEEKIT_ONPREM_HOSTS, and disables outbound internet tools (convert_url,fetch,search).TWEEKIT_REQUIRE_LOCAL_MINER_FOR_FILES=1: Restrictsconvert_fileso local workspace files can only be converted when connected to a localhost or on-premises Miner Renderery.TWEEKIT_MINER_NOAUTH=1: Permits localstdioconnections to a loopbackNOAUTH;NODBMiner Renderery without requiring cloud portal API credentials.- Runtime Introspection (
config://tweekit-runtime-mode): MCP resource returning JSON metadata (execution_mode:"onprem-miner-renderery"vs"cloud-cgn",local_only_zero_egress,local_files_enabled).
See the full architecture and last-mile installation specification in docs/onprem-miner-renderery-spec.md and our defense-in-depth security controls in SECURITY.md.
MCP Installation
You can run the TweekIT MCP server locally or connect to a hosted instance. The local option is recommended for development and testing.
Local Installation
git clone https://github.com/equilibrium-team/tweekit-mcp.git
cd tweekit-mcp
pip install -r requirements.txt
Run the server:
uv run server.py
The server will listen on:
http://localhost:8080
Config Options
Local Development Mode
By default, server.py will bind to localhost on port 8080. You can override the port with the PORT environment variable:
PORT=9090 uv run server.py
Cloud Run Deployments
Use the provided script to build and deploy containerized stage or production services on Google Cloud Run.
Prerequisites:
- Google Cloud project access (
tweekitmcp-a26b6by default). gcloudCLI authenticated (gcloud auth login) and, if your standard config directory is read-only, a writable export such asexport CLOUDSDK_CONFIG="$(pwd)/.gcloud-config".
Run the container locally with Docker (builds image tweekit-mcp-local:dev and binds to 8080):
bash scripts/deploy_cloud_run.sh local --version dev
Deploy to staging (creates/updates the tweekit-mcp-stage service):
bash scripts/deploy_cloud_run.sh stage --version 1.8.1
Deploy to production (updates the tweekit-mcp service):
bash scripts/deploy_cloud_run.sh prod --version 1.8.1
Pass --project <PROJECT_ID> if you need to override the active gcloud configuration, and --env-file <path> to supply Cloud Run environment variables in YAML format.
The helper scripts never bundle credentials. Always provide your own Google Cloud project, region, and secret sources when running them; Equilibrium’s staging/prod keys live in managed secret stores and are intentionally excluded from this repository.
Python dependencies for Firebase Functions are resolved at deploy time.
scripts/deploy_firebase.shwill vendor them intofunctions/packages/locally (a gitignored directory) so the repo stays lightweight.
Client Compatibility
- Claude Desktop / claude.ai: Compatible (HTTP transport;
.mcpbbundle available). - ChatGPT (OpenAI): Compatible (HTTP transport; plugin proxy available).
- OpenAI Codex CLI: Compatible (
--mcp-serverflag or~/.codex/config.yaml). - Cursor: Compatible (HTTP transport via
~/.cursor/mcp.json). - Continue (VS Code / JetBrains): Compatible (Streamable HTTP via
~/.continue/config.json). - Windsurf / Sourcegraph Cody: Compatible (HTTP MCP endpoints via beta settings).
Examples
-
Claude Desktop (config snippet):
{ "mcpServers": { "tweekit": { "transport": { "type": "http", "url": "https://mcp.tweekit.io/mcp" } } } } -
Claude tool call (arguments example — credentials supplied via transport headers or env vars):
{ "name": "convert", "arguments": { "inext": "png", "outfmt": "webp", "blob": "<base64>", "width": 300, "height": 300 } } -
OpenAI (Node, using MCP TypeScript client):
import { Client } from "@modelcontextprotocol/sdk/client"; import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js"; const transport = new StreamableHTTPClientTransport(new URL("https://mcp.tweekit.io/mcp"), { requestInit: { headers: { ApiKey: process.env.TWEEKIT_API_KEY ?? "", ApiSecret: process.env.TWEEKIT_API_SECRET ?? "", }, }, }); const client = new Client({ name: "tweekit-example", version: "1.0.0" }, { capabilities: {} }); await client.connect(transport); const tools = await client.listTools(); const res = await client.callTool({ name: "doctype", arguments: { extension: "pdf" }, }); console.log(res); -
OpenAI (Python, quick call):
import asyncio import os from fastmcp import Client from fastmcp.client.transports import StreamableHttpTransport async def main(): transport = StreamableHttpTransport( "https://mcp.tweekit.io/mcp", headers={ "ApiKey": os.environ["TWEEKIT_API_KEY"], "ApiSecret": os.environ["TWEEKIT_API_SECRET"], }, ) async with Client(transport) as c: print(await c.list_tools()) out = await c.call_tool( "convert", { "inext": "png", "outfmt": "txt", "blob": "<base64>", }, ) print(out) asyncio.run(main())
Quickstart (Claude Desktop)
There are two ways to use Tweekit inside Claude Desktop:
- Hosted HTTP server – point Claude at
https://mcp.tweekit.io/mcpvia Settings → Connectors → Advanced → Developer Mode → “Add HTTP server.” - Local bundle (
.mcpb) – install the packaged stdio server so you can work offline or ship curated builds.
We maintain the local bundle with scripts/build_claude_bundle.py and the guided helper scripts/configure_claude_desktop.py. The helper walks through version selection, bundle generation, dxt manifest validation, and finally reminds you to import the resulting dist/tweekit-claude.mcpb. It also reiterates that our production Tier 1 headless nodes run on Equilibrium-owned CPUcoin enterprise miners inside SAS 70 / ISO 27001 compliant data centers—the local bundle mirrors the same feature set.
After connecting (either method):
- Ask: “List supported input types via doctype.”
- Or call
convertwith a base64 blob (credentials come from your configured headers or env vars):
{
"name": "convert",
"arguments": { "inext": "png", "outfmt": "webp", "blob": "<base64>" }
}
Quickstart (ChatGPT MCP)
If your ChatGPT environment supports MCP tools, add an HTTP MCP server pointing to the public endpoint with ApiKey and ApiSecret headers.
- Configure server: URL
https://mcp.tweekit.io/mcp(HTTP transport) - Use tools in a chat:
- doctype
{ "name": "doctype", "arguments": { "extension": "pdf" } }
- convert
{
"name": "convert",
"arguments": {
"inext": "png",
"outfmt": "txt",
"blob": "<base64>"
}
}
Quickstart (Cursor IDE)
Cursor loads MCP configuration from ~/.cursor/mcp.json. To enable the hosted TweekIT server, merge the snippet below (also published at configs/cursor-mcp.json) into your existing file:
{
"mcpServers": {
"tweekit": {
"type": "http",
"url": "https://mcp.tweekit.io/mcp",
"headers": {
"ApiKey": "${TWEEKIT_API_KEY}",
"ApiSecret": "${TWEEKIT_API_SECRET}"
}
}
}
}
- Copy
.env.exampleto.env, fill inTWEEKIT_API_KEY/TWEEKIT_API_SECRET, thensource .env(or export the variables manually).
- Restart Cursor and run “List tools” to confirm
version,doctype,convert, andconvert_urlare available. - Optional: host the JSON internally and surface an “Add to Cursor” deep-link for teammates.
Quickstart (Continue IDE)
Continue (VS Code / JetBrains) stores MCP servers in ~/.continue/config.json. Add the TweekIT server under mcpServers (ready-to-copy JSON at configs/continue-mcp.json):
{
"mcpServers": {
"tweekit": {
"type": "streamable-http",
"url": "https://mcp.tweekit.io/mcp",
"headers": {
"ApiKey": "${TWEEKIT_API_KEY}",
"ApiSecret": "${TWEEKIT_API_SECRET}"
}
}
}
}
After saving the configuration:
- Reload Continue so it picks up the new MCP entry.
- Use the Continue “Tools” panel to call
doctype,convert, orconvert_urlwith the appropriate parameters. - Document workspace-specific credential steps so teammates can connect quickly.
Quickstart (OpenAI Codex CLI)
OpenAI Codex CLI (@openai/codex) supports MCP servers as first-class plugins. There are two ways to add TweekIT:
Option A — One-liner easy plugin (no config file needed):
export TWEEKIT_API_KEY=your_key TWEEKIT_API_SECRET=your_secret
codex --mcp-server tweekit:https://mcp.tweekit.io/mcp
Pass the --mcp-server flag each session, or alias it in your shell profile for permanent access.
Option B — Persistent config (~/.codex/config.yaml):
Merge the entry below (also at configs/codex-mcp.yaml) into your Codex config:
mcpServers:
tweekit:
type: http
url: https://mcp.tweekit.io/mcp
headers:
ApiKey: “${TWEEKIT_API_KEY}”
ApiSecret: “${TWEEKIT_API_SECRET}”
After connecting via either method:
- Ask: “List supported input types via TweekIT doctype.”
- Or: “Use TweekIT to convert this DOC file to PDF so you can read it.”
- Or: “Convert the image at
<url>to WebP at 800×600.”
Install Codex CLI:
npm install -g @openai/codex
Rate Limits and Pricing
TweekIT offers a generous free tier so you can explore and integrate without cost.
Free Tier
- 10,000 API calls included at no charge.
- Full access to all core upload, preview, download, and delete capabilities.
- No credit card required to start.
TweekIT MCP questions
How do I add TweekIT MCP to Claude Code?
Run claude mcp add --transport http mcp https://mcp.tweekit.io/mcp/ in your terminal. Add --scope user to make it available in every project.
How do I add TweekIT MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
What tools does TweekIT MCP provide?
It exposes 14 tools, including extension, inext, outfmt, blob, noRasterize.
How many file types does TweekIT support?
TweekIT supports over 400 different file types for ingestion and conversion.
How do I authenticate with the TweekIT MCP server?
Authentication is handled using an API Key and Secret via MCP client headers or environment variables.
Is there a way to check file compatibility without a network roundtrip?
Yes, the Python SDK includes a preflight_check function for instant local compatibility checks.