fix(agents): classify Cloudflare/CDN HTML error pages as transport failures by stainlu · Pull Request #67642 · openclaw/openclaw

stainlu · 2026-04-16T11:33:25Z

Summary

Problem: When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This causes incorrect failover behavior (profile rotation instead of clean retry/fallback) and leaves the TUI stuck.
Root cause: classifyFailoverSignal runs text-pattern classifiers on raw error messages without first checking whether the message is an HTML page. HTML error pages from CDNs like Cloudflare often contain rate-limit or error keywords in their human-readable body text, which pattern matchers incorrectly classify as structured API errors. Additionally, classifyProviderRuntimeFailureKind only checked for HTML responses on status 403 (auth_html_403), missing non-403 HTML pages entirely.
Fix:
1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly.
2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning a new "upstream_html" kind for non-403 statuses with a clear user-facing message: "The provider returned an HTML error page instead of an API response."
3. Regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML preservation, and JSON rate-limit correctness.

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

Fixes [Bug]: openai-codex/gpt-5.4 returns Cloudflare HTML and gets misclassified as rate_limit / DNS, leaving TUI stuck #67517

greptile-apps · 2026-04-16T11:36:49Z

Greptile Summary

Fixes a real misclassification bug where CDN/Cloudflare HTML error pages (502, 503, 520–524) containing text like "Rate limit exceeded" were incorrectly classified as structured API rate-limit errors, causing wrong failover behavior and a stuck TUI. The fix adds an HTML short-circuit in classifyFailoverSignal before pattern matchers run, and generalizes the HTML detection in classifyProviderRuntimeFailureKind to cover non-403 statuses with a new "upstream_html" kind and clear user-facing message. Regression tests are included for the key scenarios.

Confidence Score: 5/5

Safe to merge — the fix is correct, targeted, and well-tested for the primary regression scenarios.

Both changed functions are correctly updated, the new upstream_html kind and user-facing message are properly placed, and the key regressions are all tested. The only gap is a missing test documenting the intentional behavior change for 403 HTML in classifyFailoverReason, which is P2 and does not block merging.

No files require special attention.

Prompt To Fix All With AI

This is a comment left during a code review.
Path: src/agents/pi-embedded-helpers/provider-error-patterns.test.ts
Line: 164-169

Comment:
**Consider adding a `classifyFailoverReason` test for 403 HTML**

The HTML short-circuit in `classifyFailoverSignal` now fires for 403 HTML too, changing the failover classification from the old `"auth"` path to `"timeout"`. That behavior change is intentional (CDN 403 HTML = transport failure, not API auth failure), but it's not covered by any test here. Adding a case alongside the 502/503 tests would explicitly document the intent and prevent future regressions if the ordering of guards ever shifts.

```suggestion
  it("classifies 403 HTML as timeout in failover signal (CDN block, not API auth)", () => {
    const html403 =
      "<!doctype html><html><head><title>403 Forbidden</title></head>" +
      "<body><h1>Forbidden</h1></body></html>";
    expect(classifyFailoverReason(`403 ${html403}`)).toBe("timeout");
  });
```

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "fix(agents): classify Cloudflare/CDN HTM..." | Re-trigger Greptile}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2f43127de8

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 51869c3e8f

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 195f1fbe5c

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".

…ilures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly).

@stainlu

…hanks @stainlu)

obviyus

Verified the Cloudflare/CDN HTML failure path and confirmed the classifier now treats 5xx HTML pages as transport failures instead of misreading embedded body text as API rate limits.

Maintainer follow-up: preserved auth remediation for HTML 401/403, preserved proxy remediation for HTML 407, and fixed Error:-prefixed HTML status parsing while resolving the review-thread concerns.

Local gate: pnpm test src/agents/pi-embedded-helpers/provider-error-patterns.test.ts src/agents/pi-embedded-helpers.isbillingerrormessage.test.ts src/agents/pi-embedded-helpers.formatassistanterrortext.test.ts.

obviyus · 2026-04-16T12:50:06Z

Landed on main.

Landed source commit: a456f98
Merge commit: e588e90

Thanks @stainlu.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a456f98c91

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 · 2026-04-16T12:52:23Z

+    isTransportHtmlErrorStatus(inferredStatus) &&
+    isHtmlErrorResponse(signal.message, inferredStatus)
+  ) {


Restrict HTML timeout short-circuit to transient statuses

This new early return turns any HTML response except 401/403/407 into timeout, which now bypasses the explicit HTTP-status mapping in classifyFailoverClassificationFromHttpStatus (for example, 402 billing handling at errors.ts:556-558). A response like 402 <!doctype html>... previously classified as billing now becomes timeout, which changes failover behavior (e.g., retry-limit escalation in run/failover-policy.ts only treats non-timeout reasons as escalation candidates). Narrowing the short-circuit to transport-like statuses (5xx/408/499/etc.) would avoid regressing non-transient status handling.

Useful? React with 👍 / 👎.

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

@stainlu

…hanks @stainlu) * fix(agents): classify Cloudflare/CDN HTML error pages as transport failures Fixes openclaw#67517 When a provider endpoint returns an HTML error page (e.g. Cloudflare 502/503/520-524), the pattern-based message classifiers would scan the HTML body and misinterpret embedded text like "Rate limit exceeded" as a structured rate_limit API error. This caused incorrect failover behavior (profile rotation instead of clean retry/fallback) and left the TUI stuck. Two fixes: 1. classifyFailoverSignal now short-circuits on HTML responses before running pattern matchers, returning "timeout" (transport failure) so retry/fallback handles them correctly. 2. classifyProviderRuntimeFailureKind now detects HTML errors at any status (not just 403), returning "upstream_html" for non-403 statuses with a clear user-facing message about CDN/gateway errors. Adds regression tests covering Cloudflare 502/503 HTML with embedded rate-limit text, 403 HTML (still classified as auth), and JSON rate-limit responses (still classified correctly). * fix: preserve auth and proxy HTML classification * fix: classify HTML provider error pages correctly (openclaw#67642) (thanks @stainlu) --------- Co-authored-by: Ayaan Zaidi <hi@obviy.us>

openclaw-barnacle Bot added agents Agent runtime and tooling size: S labels Apr 16, 2026

greptile-apps Bot reviewed Apr 16, 2026

View reviewed changes

Comment thread src/agents/pi-embedded-helpers/provider-error-patterns.test.ts Outdated

chatgpt-codex-connector Bot reviewed Apr 16, 2026

View reviewed changes

Comment thread src/agents/pi-embedded-helpers/errors.ts Outdated

Comment thread src/agents/pi-embedded-helpers/errors.ts Outdated

stainlu force-pushed the fix/issue-67517-cloudflare-html-misclassification branch from 2f43127 to 51869c3 Compare April 16, 2026 11:43

chatgpt-codex-connector Bot reviewed Apr 16, 2026

View reviewed changes

Comment thread src/agents/pi-embedded-helpers/errors.ts Outdated

Comment thread src/agents/pi-embedded-helpers/errors.ts Outdated

chatgpt-codex-connector Bot reviewed Apr 16, 2026

View reviewed changes

Comment thread src/agents/pi-embedded-helpers/errors.ts

obviyus self-assigned this Apr 16, 2026

stainlu and others added 2 commits April 16, 2026 18:12

fix: preserve auth and proxy HTML classification

c926890

obviyus force-pushed the fix/issue-67517-cloudflare-html-misclassification branch from 195f1fb to 0060f57 Compare April 16, 2026 12:44

fix: classify HTML provider error pages correctly (openclaw#67642) (t…

a456f98

…hanks @stainlu)

obviyus force-pushed the fix/issue-67517-cloudflare-html-misclassification branch from 0060f57 to a456f98 Compare April 16, 2026 12:48

obviyus approved these changes Apr 16, 2026

View reviewed changes

obviyus merged commit e588e90 into openclaw:main Apr 16, 2026
9 checks passed

This was referenced Apr 16, 2026

[Bug] openai-codex provider blocked by Cloudflare JS Challenge when accessed via proxy in mainland China #67670

Closed

[Bug] openai-codex provider blocked by Cloudflare JS Challenge when accessed via proxy in mainland China #67673

Closed

chatgpt-codex-connector Bot reviewed Apr 16, 2026

View reviewed changes

github-actions Bot mentioned this pull request Apr 16, 2026

📡 Upstream Digest — 2026-04-16 13:13 UTC curtismercier/openclaw-mods#589

Open

SARAMALI15792 mentioned this pull request Apr 16, 2026

fix(agents): detect Cloudflare challenge pages and provide actionable error message #67717

Closed

25 tasks

stainlu mentioned this pull request Apr 16, 2026

fix(agents): classify raw HTML error responses even without leading HTTP status prefix #67762

Closed

7 tasks

clawsweeper Bot mentioned this pull request May 1, 2026

openai-codex provider broken since 2026.4.5 — Cloudflare challenge + missing OAuth scope #68033

Closed

clawsweeper Bot mentioned this pull request May 14, 2026

Errors: rewrite 403 HTML auth pages #73635

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(agents): classify Cloudflare/CDN HTML error pages as transport failures#67642

fix(agents): classify Cloudflare/CDN HTML error pages as transport failures#67642
obviyus merged 3 commits into
openclaw:mainfrom
stainlu:fix/issue-67517-cloudflare-html-misclassification

stainlu commented Apr 16, 2026

Uh oh!

greptile-apps Bot commented Apr 16, 2026

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

obviyus left a comment

Uh oh!

Uh oh!

obviyus commented Apr 16, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Apr 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

stainlu commented Apr 16, 2026

Summary

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

Uh oh!

greptile-apps Bot commented Apr 16, 2026

Greptile Summary

Confidence Score: 5/5

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

obviyus left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

obviyus commented Apr 16, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Apr 16, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants