Use an HTTP proxy with the ValidMind Library

Published

May 2, 2026

Outbound HTTPS traffic to the ValidMind API relies on standard Python HTTP clients that honor proxy-related environment variables, which lets you work behind corporate or regional proxies without custom library configuration.

Prerequisites

1 If you have not installed the library yet, start with Install and initialize the ValidMind Library.

What do I need to know?

Underlying behavior

Proxy handling is implemented by the underlying HTTP stacks, including aiohttp and requests for ValidMind API traffic, and httpx in some third-party clients. The ValidMind Library itself does not provide a way to set a proxy on vm.init().

Precedence between ALL_PROXY and scheme-specific variables, and exact NO_PROXY matching rules, can vary by library version. If results differ from what you expect, refer to the documentation for those libraries or your HTTP client.2

Other outbound traffic

Features that generate text with OpenAI or Azure OpenAI use the Python SDKs provided by those vendors. These clients usually respect the same HTTP_PROXY, HTTPS_PROXY, and NO_PROXY values, but authentication, regional endpoints, and enterprise policies can add requirements.

Consult your provider’s documentation when something works for the ValidMind API but not for LLM calls.3

3 For example, refer to OpenAI and Azure OpenAI for service-specific networking guidance.

Step 1: Set proxy environment variables

Set the variables before you start the Python process or Jupyter kernel, so that the process inherits them when the library opens HTTP connections.

Common variables include:

Variable Purpose
HTTP_PROXY Proxy URL for http requests.
HTTPS_PROXY Proxy URL for https requests typically used for the ValidMind API.
ALL_PROXY Fallback when a scheme-specific variable is not set. Behavior depends on the client library.
NO_PROXY Comma-separated hostnames, domains, or IP ranges that should NOT use the proxy. For example: localhost, 127.0.0.1, or an internal API hostname.

Many tools and Linux environments use lowercase names, such as http_proxy or https_proxy. On Unix-like systems, Python and common HTTP clients often treat uppercase and lowercase as equivalent for these variables, but some scripts or containers only read one convention.

If your organization assigns these values automatically when you log in, confirm they are visible in the same environment where you launch Jupyter or your IDE.

Tip: Use the names your organization or image already exports, or standardize on a single case for your team.

Example for a POSIX shell

export HTTPS_PROXY=http://proxy.example.com:8080
export HTTP_PROXY=http://proxy.example.com:8080
export NO_PROXY=localhost,127.0.0.1,.corp.example.com

Step 2: Restart your notebook or Python session

If you change proxy variables after a session has already started, restart the kernel or Python interpreter, so that every library reload sees the updated environment.

Troubleshooting

  • If you see SSL or certificate verification errors when calling vm.init() behind a proxy: refer to our troubleshooting guide.4
  • If you see connection timeouts or “connection refused” when calling vm.init() or logging results: verify HTTPS_PROXY matches your proxy URL and that the host running Python can reach the proxy host and port.
  • If traffic should bypass the proxy for a specific API host: add that host or domain to NO_PROXY and restart the session.
  • If variables do not appear to apply: in a notebook, run import os and print(os.environ.get("HTTPS_PROXY")) in a fresh cell after restart; if it is empty, the variables were not set for that process (for example, they were only set in a different terminal).

What's next