fix: resolve model discovery to /v1/models regardless of baseUrl

[devin-ai-integration]

Summary

Model discovery failed for the recommended Anthropic Messages config. The extension fetched `${provider.baseUrl}/models`, but the Pi docs configure api: "anthropic-messages" with baseUrl: https://router.requesty.ai (deliberately without /v1, so Pi appends /v1/messages correctly). That made discovery hit https://router.requesty.ai/models → 404, so both startup discovery and /requesty-models-sync failed for anyone following the docs.

Fix: derive the models endpoint so it always targets /v1/models, regardless of whether baseUrl includes a version segment:

function modelsUrl(baseUrl) {
  const base = normalizeBaseUrl(baseUrl);
  return /\/v\d+$/.test(base) ? `${base}/models` : `${base}/v1/models`;
}
// discoverModels now fetches modelsUrl(provider.baseUrl)

This keeps the existing OpenAI-completions config (baseUrl: .../v1) working while fixing the Anthropic-messages config (baseUrl: ... root).

Testing

Verified live against router.requesty.ai: /v1/models → 200 (592 models), /models → 404. Ran the actual extension module with the docs' anthropic-messages config: startup registers 592 models and /requesty-models-sync writes 592 models back to models.json, preserving baseUrl/api. Both /v1 and no-/v1 base URLs resolve to https://router.requesty.ai/v1/models.

Link to Devin session: https://app.devin.ai/sessions/d91233e460d344fd9b46cee67d79bb2f
Requested by: @Thibault00

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[requesty-ditti]

Can we please add unit-tests for both normalizeBaseUrl and modelsUrl?

[requesty-ditti]

It would help to see the exact strings that are coming in and out here

[devin-ai-integration]

Added unit tests for both normalizeBaseUrl and modelsUrl in requesty.test.js (both functions are now exported), plus a npm test script using Node's built-in test runner (node --test). Both tests pass.

[devin-ai-integration]

The new tests document the exact in/out strings. Highlights:

normalizeBaseUrl:

• https://router.requesty.ai/ → https://router.requesty.ai
• https://router.requesty.ai/v1/ → https://router.requesty.ai/v1

modelsUrl:

• https://router.requesty.ai (Anthropic Messages config) → https://router.requesty.ai/v1/models
• https://router.requesty.ai/v1 (OpenAI Completions config) → https://router.requesty.ai/v1/models
• https://router.requesty.ai/v2 → https://router.requesty.ai/v2/models

So regardless of whether the configured baseUrl includes a version segment, discovery always resolves to a single /v1/models (no /v1/v1/models and no bare /models).

[tobru]

@requesty-ditti Any chance to get this merged? I'd love to use Pi with Requesty instead of with OpenRouter, but this is holding me back right now.