fix(fetch): route dynamic model-list fetches through request options

[devYRPauli]

Problem

The dynamic model-list fetchers in core/llm/fetchModels.ts (Ollama, OpenRouter, Anthropic, Gemini) call the global fetch() directly. They never go through fetchwithRequestOptions, so they ignore the user's requestOptions (proxy, custom CA / verifySsl, custom headers).

Behind an SSL-intercepting corporate proxy, or with a custom CA configured, the "Add model" dropdown fails to load these provider model lists even though normal chat requests succeed - because chat already routes through fetchwithRequestOptions while these fetchers do not.

Fix

Thread requestOptions through the four fetchers and use fetchwithRequestOptions instead of the raw fetch(). The models/fetch handler in core.ts now loads the config and passes config.requestOptions down.

This mirrors the earlier ContinueServerClient fix in #12993, which deliberately left fetchModels.ts untouched.

Notes

• No behavior change for users without requestOptions configured; it stays a no-op for them.
• The change follows the existing fetchwithRequestOptions(url, init, requestOptions) call pattern already used elsewhere in core.ts.

[chatgpt-codex-connector]

💡 Codex Review

continue/docs-site/app/[[...slug]]/page.tsx

Lines 69 to 71 in a7bce9f

	const target = redirectTo.startsWith("http")
	? redirectTo
	: `/docs${redirectTo}`;

Preserve standalone docs paths when redirecting

For legacy redirects from docs/docs.json with internal destinations such as /cli/overview -> /cli/quickstart, this constructs /docs/cli/quickstart. The standalone docs app generates pages at /cli/quickstart and other code routes links through resolveHref to strip /docs, so redirected users land on an ungenerated/404 path in the static export. Route internal redirects through resolveHref or use redirectTo as-is.

---

continue/core/llm/fetchModels.ts

Lines 247 to 251 in a7bce9f

	const llm = llmFromProviderAndOptions(provider, {
	apiKey,
	apiBase: apiBase || defaultApiBase,
	model: "",
	});

Pass request options into listModels fallback

When a provider falls through to listModels (for example the OpenAI/Mistral refresh button in Add Model), this LLM is constructed without requestOptions, so BaseLLM.fetch has no proxy, custom CA, or custom headers. Users with requestOptions configured still fail on these model-list fetches even though models/fetch loaded config.requestOptions; thread the option into this fallback constructor too.

---

continue/extensions/vscode/src/commands.ts

Lines 712 to 714 in a7bce9f

	const isValid = core.invoke("mdm/setLicenseKey", {
	licenseKey,
	});

Restore enterprise license handling

After removing the core mdm/setLicenseKey listener, this command is still contributed and calls it. In the VS Code in-process messenger, invoke returns undefined when no listener is registered, so any entered enterprise key is treated as invalid; JetBrains' AddLicenseKey action still sends the same message too. Restore the handler or remove/disable the command.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".