fix(gateway): web_search tool in responses api

[steebchen]

Summary

A user reported that web_search tools work via /v1/chat/completions but are silently ignored via /v1/responses (the model answers "I don't have real-time access" and web_search_cost stays 0).

Root cause: the Responses API handler converts the request to an internal chat completions request but dropped every non-function tool, including web_search — even though the chat completions layer supports it natively (provider grounding / native web search).

Changes

• responses.ts: pass web_search tools through to the internal chat completions request instead of dropping them. Other OpenAI-native built-ins (computer_use, code_interpreter, …) are still dropped.
• schemas.ts: accept allowed_domains/blocked_domains on the responses web_search tool for parity with chat completions (zod was silently stripping them).
• convert-chat-to-responses.ts: flatten chat-style url_citation annotations (nested under url_citation) to the Responses API annotation shape.
• convert-streaming-to-responses.ts: forward citation annotations in the streaming path (previously hardcoded to []) — accumulates delta.annotations, emits response.output_text.annotation.added events, and includes them in content_part.done/output_item.done/final output.
• chat-websearch.e2e.ts: new e2e test exercising web_search via /v1/responses, asserting output text and webSearchCost > 0 in logs.

Testing

• TEST_WEB_SEARCH=1 TEST_MODELS="google-ai-studio/gemini-3.5-flash" pnpm test:e2e — all three web search tests pass (non-streaming, responses api, streaming) against the real provider with webSearchCost > 0. (Remaining failures in that run are pre-existing gemini JSON-mode flakes on the untouched chat completions path.)
• pnpm test:unit — only pre-existing failure (api.spec.ts hybrid-escape test, fails on a clean tree too).
• pnpm build passes.

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features
  • Added support for web search through the Responses API.
  • Web search tools can now restrict results using allowed or blocked domains.
  • Web search results and citations are preserved in both standard and streaming responses.
• Improvements
  • URL citation data is returned in a consistent format across response types.
  • Streaming responses now emit citation events as annotations become available.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[coderabbitai]

Warning

Review limit reached

@steebchen, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 27 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: cfab40cf-6d7e-4086-afb7-0debe80e94ee

📥 Commits

Reviewing files that changed from the base of the PR and between 41096ce and cd7ddae.

📒 Files selected for processing (5)

• apps/gateway/src/chat-websearch.e2e.ts
• apps/gateway/src/responses/responses.ts
• apps/gateway/src/responses/schemas.ts
• apps/gateway/src/responses/tools/convert-chat-to-responses.ts
• apps/gateway/src/responses/tools/convert-streaming-to-responses.ts

Walkthrough

The Responses API now forwards web_search tools with domain filters, validates non-streaming web search responses and costs, and normalizes URL citations in both non-streaming and streaming output conversions.

Changes

Responses web search and annotations

Layer / File(s)	Summary
Web search tool passthrough and coverage apps/gateway/src/responses/schemas.ts, apps/gateway/src/responses/responses.ts, apps/gateway/src/chat-websearch.e2e.ts	The Responses schema accepts allowed_domains and blocked_domains, the handler forwards web_search, and end-to-end coverage validates output and webSearchCost.
Non-streaming annotation normalization apps/gateway/src/responses/tools/convert-chat-to-responses.ts	Nested chat-completions url_citation fields are flattened before being emitted in Responses output.
Streaming annotation propagation apps/gateway/src/responses/tools/convert-streaming-to-responses.ts	Streaming annotations are normalized, accumulated, emitted as events, and included in completed output messages.

Estimated code review effort: 3 (Moderate) | ~25 minutes

Sequence Diagram(s)

sequenceDiagram
  participant Client
  participant ResponsesEndpoint
  participant ChatCompletions
  participant RequestLogs
  Client->>ResponsesEndpoint: POST /v1/responses with web_search
  ResponsesEndpoint->>ChatCompletions: Forward web_search tool
  ChatCompletions-->>ResponsesEndpoint: Return search response
  ResponsesEndpoint->>RequestLogs: Record webSearchCost
  ResponsesEndpoint-->>Client: Return Responses output

Loading

sequenceDiagram
  participant ChatCompletionsStream
  participant StreamConverter
  participant StreamingState
  participant ResponsesClient
  ChatCompletionsStream->>StreamConverter: Send delta annotations
  StreamConverter->>StreamingState: Normalize and accumulate annotations
  StreamConverter-->>ResponsesClient: Emit annotation.added event
  StreamingState-->>ResponsesClient: Include annotations in completed output

Loading

Possibly related PRs

• theopenco/llmgateway#1983: Modifies the same response conversion function for assistant output handling.
• theopenco/llmgateway#2287: Also updates streaming and non-streaming Responses annotation output.
• theopenco/llmgateway#2702: Extends web search tool fields through the request conversion path.

Suggested reviewers: smakosh, RATCHAW

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly reflects the main change: fixing web_search support in the gateway Responses API.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix-web-search-responses-api

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}