fix(typescript): handle undici timeout errors and retry network failures

[devin-ai-integration]

Description

Fixes an issue where Node.js's built-in fetch (backed by undici) has a hard 300s headersTimeout that fires before the SDK's timeoutInSeconds, causing generic "fetch failed" errors on long-running requests. Reported upstream at elevenlabs/elevenlabs-js#379.

Changes Made

1. Detect undici timeout errors (Fetcher.ts)

• Added isTimeoutError() helper that checks for undici-specific error codes (UND_ERR_HEADERS_TIMEOUT, UND_ERR_BODY_TIMEOUT, ETIMEDOUT) in the error cause chain
• These are now classified as reason: "timeout" instead of reason: "unknown", allowing SDK users to handle them with proper timeout error handling

2. Surface error cause chain (Fetcher.ts)

• Added getErrorMessageWithCause() helper that walks the error.cause chain to build a descriptive error message including underlying error codes
• Before: errorMessage: "fetch failed" — opaque, unhelpful
• After: errorMessage: "fetch failed -> Headers Timeout Error [UND_ERR_HEADERS_TIMEOUT]" — actionable

3. Retry on network-level failures (requestWithRetries.ts)

• Restructured the retry loop to catch thrown errors from fetch (not just HTTP status codes)
• Only transient network errors (e.g. ECONNRESET, ECONNREFUSED, EPIPE) are retried with exponential backoff up to maxRetries
• AbortError (SDK timeout / user abort) and undici timeout errors (UND_ERR_HEADERS_TIMEOUT, UND_ERR_BODY_TIMEOUT, ETIMEDOUT) propagate immediately without retries via isRetryableError() filter
• Previously, thrown errors from the fetch call bypassed retry logic entirely — the request failed immediately regardless of maxRetries
• Backward-compatible: existing behavior for HTTP-status-based retries (408, 429, 5xx) is unchanged

Seed snapshots

• Updated all 211 Fetcher.ts and requestWithRetries.ts seed fixtures across seed/ts-sdk/

Testing

• pnpm check (biome lint) passes
• pnpm format passes
• Seed snapshot files updated to match source templates

Link to Devin session: https://app.devin.ai/sessions/234b24e109924874ab8af5e9b07f397e

---

[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 and CI monitoring

[claude]

Claude Code Review

This repository is configured for manual code reviews. Comment @claude review to trigger a review and subscribe this PR to future pushes, or @claude review once for a one-time review.

_{Tip: disable this comment in your organization's Code Review settings.}

[github-actions]

SDK Generation Benchmark Results

Comparing PR branch against median of 5 nightly run(s) on main (latest: 2026-04-23T04:59:11Z).

Full benchmark table (click to expand)

Generator	Spec	main (generator)	main (E2E)	PR (generator)	Delta
ts-sdk	square	74s (n=5)	86s (n=5)	68s	-6s (-8.1%)

main (generator): generator-only time via --skip-scripts (includes Docker image build, container startup, IR parsing, and code generation — this is the same Docker-based flow customers use via fern generate). main (E2E): full customer-observable time including build/test scripts (nightly baseline, informational). Delta is computed against generator-only baseline.
⚠️ = generation exited with a non-zero exit code (timing may not reflect a successful run).
Baseline from nightly runs on main (latest: 2026-04-23T04:59:11Z). Trigger benchmark-baseline to refresh.
Last updated: 2026-04-29 13:46 UTC