Altimate Base: 413 HTML error body shows generic message; no client-side header timeout

[anandgupta42]

Summary

Two client-side bugs found during live-prod E2E of the v0.11.0-beta.2 release, both in the Altimate Base free-tier code path.

Bug 1 — 413 oversize error shows a generic message instead of the friendly one

FreeTier.describeRequestTooLarge (packages/opencode/src/altimate/free/client.ts) tries to JSON.parse the response body and returns undefined if parsing throws.

In production, oversized requests (roughly >1 MiB) are rejected by the gateway's edge proxy with a raw HTML error page, not a JSON body. JSON.parse throws on that HTML, the function returns undefined, and the user is shown a generic fallback error instead of the friendly "your request is too large — start a new session or shorten it" guidance that fires when the body is parseable JSON.

Expected: on any HTTP 413 from the Altimate Base provider, the user should see the friendly "request too large" message regardless of whether the response body is JSON or HTML.

Actual: only a parseable JSON body with the expected shape produces the friendly message; an HTML/unparseable body falls through to the generic error.

Bug 2 — Altimate Base provider has no client-side header timeout

The altimate-free provider loader (packages/opencode/src/provider/provider.ts) does not set a headerTimeout option, unlike the openai loader in the same file which sets one by default.

If the gateway hangs before sending response headers, a request through the Altimate Base provider has no client-side timeout and can hang indefinitely — the CLI never surfaces an error or times out on its own.

Expected: the Altimate Base provider should time out client-side (matching the existing OpenAI-loader default) if the gateway doesn't respond in time.

Actual: no timeout is configured, so a hung gateway request hangs the CLI.

How found

Both were found via live-prod end-to-end testing of the v0.11.0-beta.2 release.