workflow-engine: persist step error for exhaustion + regression test

[NathanFlurry]

Description

Please include a summary of the changes and the related issue. Please also include relevant motivation and context.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

Please describe the tests that you ran to verify your changes.

Checklist:

• My code follows the style guidelines of this project
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[railway-app]

🚅 Deployed to the rivet-pr-4185 environment in rivet-frontend

Service	Status	Web	Updated (UTC)
website	😴 Sleeping (View Logs)	Web	Feb 13, 2026 at 8:51 pm
frontend-inspector	❌ Build Failed (View Logs)	Web	Feb 12, 2026 at 9:02 am
frontend-cloud	❌ Build Failed (View Logs)	Web	Feb 12, 2026 at 9:02 am
mcp-hub	✅ Success (View Logs)	Web	Feb 12, 2026 at 9:02 am
ladle	❌ Build Failed (View Logs)	Web	Feb 12, 2026 at 9:01 am

[NathanFlurry]

• workflow-engine: persist step error for exhaustion + regression test #4185 👈 (View in Graphite)
• main

---

How to use the Graphite Merge Queue

Add the label merge-queue to this PR to add it to the merge queue.

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

[claude]

Code Review

Summary

This PR fixes a bug where step exhaustion errors could lose the underlying error information when driver implementations crash or partially persist state between attempts. The fix ensures that metadata.error is persisted as a fallback source of truth.

Strengths

• Good defensive programming - The fallback logic at line 395 (stepData.error ?? metadata.error) handles edge cases where history entries may be incomplete due to partial writes or crashes.

• Comprehensive test coverage - The StripStepHistoryErrorDriver test helper effectively simulates the failure scenario and validates both yield and live modes.

• Consistent error persistence - Now all error paths (timeout, critical, rollback, and regular failures) persist metadata.error, ensuring the fallback always has data.

• Clear commenting - Lines 392-394 explain the rationale for the fallback approach.

Issues Found

Critical: Indentation inconsistency (context.ts:446-502)

The success and error handling blocks have incorrect indentation. All code from line 446 onwards should be dedented by one tab to align with the surrounding code structure. This appears to be a formatting artifact.

Minor: metadata.error clearing on success (line 451)

While the PR adds metadata.error = undefined at line 451 when a step completes successfully, this is good practice but should be verified.

Recommendations

1. Fix indentation - Run formatting on context.ts lines 446-502 to correct the extra indentation throughout the success and error blocks.

2. Test edge case - Consider adding a test where step fails once (metadata.error set), succeeds on retry (metadata.error cleared), and verify metadata.error is undefined after success.

3. Documentation - Consider adding a comment near the EntryMetadata interface explaining that metadata.error serves as a fallback for stepData.error in crash scenarios.

Security and Performance

No security concerns or performance impact identified.

Overall Assessment

Approve with changes required - The logic fix is sound and the test coverage is excellent, but the indentation issue must be corrected before merging.