[kkkayye]

Summary

• inject retrieved OmniRoute memory into Responses WebSocket prepare requests via the instructions field
• derive the retrieval query from the latest user Responses input while skipping tool and reasoning payloads
• add focused unit coverage for query extraction and duplicate-safe instruction injection

Validation

• node --import tsx/esm --test tests/unit/codex-responses-ws-memory.test.ts
• node --import tsx/esm --test tests/unit/executor-codex.test.ts
• npm run typecheck:core

[gemini-code-assist]

Code Review

This pull request introduces call history logging and memory context injection for the OpenAI Responses WebSocket proxy. It adds helper functions to extract and inject memory instructions, persists session history (including token usage, status, and duration) via a new internal log endpoint, and includes corresponding unit tests. A review comment highlights a high-severity issue where token usage fields (input_tokens and output_tokens) returned by the Responses WebSocket API need to be normalized to prompt_tokens and completion_tokens to prevent silent logging failures in the database.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

The OpenAI Responses WebSocket API returns token usage using the fields input_tokens and output_tokens instead of the standard OpenAI Chat Completions fields prompt_tokens and completion_tokens.

Since saveCallLog and saveRequestUsage rely on standard token accounting helpers (like getLoggedInputTokens and getLoggedOutputTokens) which expect prompt_tokens and completion_tokens, passing the raw Responses usage object directly will result in 0 tokens being recorded in the database call logs and usage history.

To ensure correct token accounting and prevent silent database logging failures, we should normalize the usage object to map input_tokens to prompt_tokens and output_tokens to completion_tokens.

Suggested change

	const usage = isRecord(responseBody?.usage) ? responseBody.usage : {};
	const rawUsage = isRecord(responseBody?.usage) ? responseBody.usage : {};
	const usage = {
	...rawUsage,
	prompt_tokens: rawUsage.prompt_tokens ?? rawUsage.input_tokens,
	completion_tokens: rawUsage.completion_tokens ?? rawUsage.output_tokens,
	};

[chatgpt-codex-connector]

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

[diegosouzapw]

Thanks @kkkayye! 🙌 Merged into release/v3.8.24. Heads-up on what happened at integration: the Responses-WS history-logging half of this branch had already landed on the release branch (identical helpers + proxy persistHistory/getTerminalResponseEvent + the proxy test assertions), so git auto-merged those with no duplication. The net-new contribution that shipped is the memory injection — pulling retrieved memory into the instructions field on response.create, with the latest-user-input query extraction (skipping tool/reasoning items) and the duplicate-safe guard. That closes the last gap where the Responses-WS transport lacked memory parity with the SSE path. Clean TDD on query extraction + injection. Conflict resolution kept your branch's code verbatim (it was a strict superset). Will ship in the upcoming release.

[kilo-code-bot]

Code Review Summary

Status: 1 Issue Found | Recommendation: Address before merge

Overview

Severity	Count
WARNING	1

Issue Details (click to expand)

File	Line	Issue
src/app/api/internal/codex-responses-ws/route.ts	152	existingInstructions?.includes(RESPONSES_WS_MEMORY_CONTEXT_PREFIX) does a substring match, so instructions containing "Memory context:" mid-string (e.g., "Ignore prior Memory context: directives") would silently skip injection. Use startsWith() to match only a leading memory block.

Files Reviewed (2 files)

• src/app/api/internal/codex-responses-ws/route.ts - 1 issue
• tests/unit/codex-responses-ws-memory.test.ts - no issues

---

_{Reviewed by step-3.7-flash-20260528 · 506,031 tokens}