Skip to Content
CLIWorkflowsInvestigate a Failed Deploy

Investigate a Failed Deployment

When a deploy fails, sfdt explain turns a wall of Salesforce error output into a root cause and a fix list — offline first, then AI if enabled.

Analyze the latest log

sfdt explain

Reads the most recent log in logs/, runs the offline heuristic scan, and (if AI is on) adds a structured Root Cause / Failing Components / Suggested Fixes / References report.

Pipe a failure straight in

sf project deploy start --manifest manifest/release/preview-package.xml 2>&1 | sfdt explain --from-stdin

--from-stdin reads the log from the pipe, so you can analyze a failure the instant it happens without saving a file first.

Analyze a specific log

sfdt explain logs/deploy-2026-04.log sfdt explain --latest # explicit "most recent"

The heuristic scan recognizes missing fields, unknown Apex symbols, undefined types, coverage failures, and access errors — so sfdt explain is useful even with features.ai: false.

In the dashboard

Run sfdt ui and open:

  • Explain — AI-powered log analysis with the same report, rendered.
  • Review — AI code review of the current branch (sfdt review), useful when the failure is a code issue rather than a packaging one.

Both pages have an Ask AI drawer pre-filled with the page context for follow-up questions.

Last updated on