[WRONG BRANCH] fix(web-search): bound xAI error response bodies

[luvs01]

Motivation

• The xAI sidecar executor read non-OK upstream response bodies with res.text(), which can materialize an arbitrarily large untrusted body and risk memory exhaustion.

Description

• Replace the unbounded res.text() read in runXaiWebSearch with the repository's bounded reader readBoundedResponseBytes and cap the read at MAX_SIDECAR_RESPONSE_BYTES so oversized error bodies are not retained in memory.
• On an oversized non-OK body the code now cancels the upstream stream and returns a safe, short error detail rather than decoding the full payload, while preserving entitlement/status messaging and existing timeout/abort behaviour.
• Preserve the successful-path SSE parsing in parseXaiResponsesSSE unchanged so normal streaming remains byte-bounded by the existing parser limits.
• Add a focused regression in tests/xai-web-search.test.ts that simulates a large non-OK response stream and asserts the upstream is canceled and the returned error is the bounded message.

Testing

• Ran the focused suite with node_modules/bun/bin/bun.exe test tests/xai-web-search.test.ts and the file's tests passed (14 passed, 0 failed).
• Ran bun x tsc --noEmit (typecheck) and bun run privacy:scan, both succeeded.
• Ran a broader bun run test; unrelated pre-existing failures surfaced in other suites during that full run, but the focused xAI web-search tests and typecheck/privacy scan that verify this change are green.

Files changed: src/web-search/xai-executor.ts, tests/xai-web-search.test.ts.

---

Codex Task

[github-actions]

✅ Deterministic PR hygiene checks passed.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: f2a9dc94-e1b8-4d46-aab9-ecbd1e31cbd6

---

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.}

[github-actions]

⏳ DRAFT

• wrong target branch (main); retarget to dev.

What to do

• Retarget this PR to dev — all contributions go to dev.

Its title has been prefixed with [WRONG BRANCH].
This pull request was already a draft. Its draft status will be preserved after every issue above is resolved.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4139d7dd3e

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".

[chatgpt-codex-connector]

Preserve HTTP status when bounded body reading fails

When xAI has already returned non-OK headers but its body stream subsequently rejects—for example, because the connection resets while reading—the new await propagates to the outer catch and replaces the known HTTP status and any 401/403 entitlement hint with a generic body-read/connect error. The previous res.text().catch(() => "") retained the status-only error in this scenario; catch read failures within the non-OK branch and fall back to that status-based representation while still propagating intentional aborts.

AGENTS.md reference: src/AGENTS.md:L17-L17

Useful? React with 👍 / 👎.