Development

[MitulShah1]

No description provided.

[chatgpt-codex-connector]

💡 Codex Review

model-catalog/scrape/api/compatible.go

Lines 90 to 92 in 9c70b63

	var resp openAICompatibleListResponse
	if err := json.Unmarshal(data, &resp); err != nil {
	return nil, fmt.Errorf("parse: %w", err)

Handle Mistral list response format in parser

NewMistral points at https://api.mistral.ai/v1/models, but this parser only accepts an OpenAI-style object with a data field. Mistral’s models endpoint returns a top-level JSON array, so unmarshalling fails and ferrocat freshness will always emit a warning and skip Mistral coverage whenever MISTRAL_API_KEY is set. Supporting both array and {data:[...]} shapes (or using a Mistral-specific parser) is needed for the new check to work.

---

model-catalog/catalog/prune.go

Lines 42 to 46 in 9c70b63

	pruned++
	if dryRun {
	continue
	}
	if err := os.Remove(filepath.Clean(path)); err != nil {

Prevent pruning bases that still have extends dependents

This deletion path removes sunsetted model files without checking whether other YAML entries still extends them. In the current catalog there are wrapper models whose own sunset_date is null while their base has a dated sunset; once the base crosses the cutoff, monthly prune will delete only the base and the subsequent build fails in ResolveExtends (extends ... which does not exist). Pruning should either cascade to wrappers or skip bases with live dependents.

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".