fix(logs): logging with error issues for model costs

[icecrasher321]

Summary

Async Executions got removed in new executor

Schedules, Webhooks, API executions are not tracking LLM cost when there's a downstream failure

Schedules execution errors — always surface as workflow error without trace psan

Type of Change

• Bug fix
• New feature
• Breaking change
• Documentation
• Other: ___________

Testing

Tested manually

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Preview	Comments	Updated (UTC)
docs	Skipped			Dec 3, 2025 4:50am

[greptile-apps]

Greptile Overview

Greptile Summary

Restores async execution support and fixes critical cost tracking issues for scheduled workflows.

• Async Execution: Re-added handleAsyncExecution function to queue workflows via Trigger.dev when X-Execution-Mode: async header is set
• Schedule Error Tracking: Fixed schedule executions to collect trace spans even on failure by extracting executionResult from errors and calling buildTraceSpans before logging
• Cost Calculation on Error: Modified completeWithError to calculate cost summary from provided trace spans instead of defaulting to zero cost, ensuring LLM costs are tracked when workflows fail downstream
• Token Field Compatibility: Updated token counting to handle both new (input/output) and legacy (prompt/completion) field names using nullish coalescing operators

The changes ensure proper cost tracking for failed executions and restore the ability to run workflows asynchronously for API, webhook, and schedule triggers.

Confidence Score: 5/5

• This PR is safe to merge with no blocking issues identified
• All changes are bug fixes that restore missing functionality and fix critical cost tracking issues. The implementation properly handles edge cases, uses defensive programming with nullish coalescing, and follows established patterns from the codebase
• No files require special attention

Important Files Changed

File Analysis

Filename	Score	Overview
apps/sim/app/api/workflows/[id]/execute/route.ts	5/5	Restored async execution functionality by adding handleAsyncExecution function and checking X-Execution-Mode header
apps/sim/background/schedule-execution.ts	5/5	Fixed trace span collection on schedule failures by extracting executionResult and building trace spans before logging errors
apps/sim/lib/logs/execution/logging-factory.ts	5/5	Fixed token counting to handle both input/output and legacy prompt/completion token field names using nullish coalescing
apps/sim/lib/logs/execution/logging-session.ts	5/5	Fixed cost tracking when workflows error by calculating cost summary from provided trace spans instead of always using zero cost

Sequence Diagram

sequenceDiagram
participant Client
participant API as Workflow Execute API
participant TriggerDev as Trigger.dev
participant Executor as Workflow Executor
participant Logger as Logging Session
participant TraceSpans as Trace Spans Builder
participant DB as Database
Note over Client,DB: Async Execution Flow (Restored)
Client->>API: POST /api/workflows/[id]/execute<br/>X-Execution-Mode: async
API->>API: Check isAsyncMode header
API->>TriggerDev: tasks.trigger('workflow-execution')
TriggerDev-->>API: Return handle with jobId
API-->>Client: 202 Accepted with jobId
Note over Client,DB: Schedule Execution with Error
TriggerDev->>Executor: Execute scheduled workflow
Executor->>Logger: safeStart()
Executor-->>TriggerDev: Execution fails with earlyError
TriggerDev->>TraceSpans: buildTraceSpans(executionResult)
TraceSpans-->>TriggerDev: Return traceSpans with cost data
TriggerDev->>Logger: safeCompleteWithError({traceSpans})
Logger->>Logger: calculateCostSummary(traceSpans)
Logger->>DB: Save execution log with costs
Note over Client,DB: Cost Tracking Fix
Logger->>Logger: Check if traceSpans provided
alt Has trace spans
Logger->>Logger: calculateCostSummary(traceSpans)
Note over Logger: Uses input/output or<br/>prompt/completion tokens
else No trace spans
Logger->>Logger: Use BASE_EXECUTION_CHARGE only
end
Logger->>DB: Save with calculated costs

Loading

[greptile-apps]

_{2 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}