debug-flow
bởi microsoft
Gỡ lỗi một lần chạy luồng Power Automate bị lỗi. Sử dụng khi luồng bị lỗi, có lỗi hoặc người dùng muốn khắc phục sự cố một lần chạy.
npx skills add https://github.com/microsoft/power-platform-skills --skill debug-flowDebug a Failed Flow Run
You are helping the user debug a failed Power Automate flow run.
Tools
This skill uses the FlowAgent MCP tools. Clients surface them with a
client-specific prefix — mcp__flowagent__<tool> (Claude Code) or
flowagent-<tool> (Copilot CLI) — so they're referred to by bare name below.
If MCP tools aren't available, run /setup to wire the FlowAgent MCP server.
| Tool | Purpose |
|---|---|
list_flows | Find flows by name (use name param for search) |
get_run_history | Get recent runs for a flow |
diagnose_run | One-shot: classify failed actions with remediations |
get_run_details | Get details for a specific run |
get_run_actions | Get action-level execution trace |
get_run_action_repetitions | Iteration-level detail for a loop (which iteration failed — pass an action inside the loop) |
get_flow | Get flow definition for context |
get_operation_details | Confirm the correct action type / parameters |
edit_flow | Apply a surgical fix to one action/parameter |
update_flow | Replace the whole definition (large rewrites) |
run_flow | Re-run after fix (use wait: true to see result) |
resubmit_run / cancel_run | Resubmit a fixed run / cancel a stuck one |
Steps
-
Identify the flow and run
- Parse
$ARGUMENTSfor flow ID and optional run ID. - If no flow ID, call
list_flowswithnameparam to search. Ask user to pick if ambiguous. - If no run ID, call
get_run_historyand find the most recent failed run. - Present recent runs in a table: Run ID | Status | Start Time | Error
- Parse
-
Fast triage with
diagnose_run- Call
diagnose_runfor the run — it returns the failed/timed-out actions already classified with a remediation each. Use this as the starting point. - For deeper analysis, also fetch (in parallel):
get_run_actions(full trace) andget_flow(definition context). For a failed loop, callget_run_action_repetitionson an action inside the loop to find the failing iteration.
- Call
-
Analyze failures
- Identify actions with status != "Succeeded"
- Trace the
runAfterdependency chain to separate root cause from cascading failures:- Root cause: action whose dependencies all Succeeded but it failed
- Cascading: actions skipped because a dependency failed
- Report root cause actions with: name, type, error code, error message
-
Root cause classification
- Connection errors (
AuthorizationFailed,ConnectionNotFound,InvokerConnectionOverrideFailed): Suggest re-auth or fix connection source to Embedded - Expression errors (
ExpressionEvaluationFailed,InvalidTemplate): Show the expression, explain what's wrong, suggest fix - API/External errors (401/403/404/429/500+): Explain the HTTP error, check connector status
- Parameter errors (
WorkflowOperationParametersRuntimeMissingValue): Missing/empty required parameter. Add null guard:@if(empty(...), 'default', ...) - Timeout (
ActionTimedOut): Suggest retry policy or splitting the operation - Type mismatch (
InvalidOpenApiConnectionOperationType): Wrong action type. Callget_operation_detailsto find correct type.
- Connection errors (
-
Suggest fix
- Provide a specific fix with code/expression changes.
- If the fix is a definition change, offer to apply it with
edit_flow(surgical, one action/parameter) — fall back toupdate_flowonly for large rewrites.
-
Re-test
- Offer to call
run_flowwithwait: trueto verify the fix works (orresubmit_runto retry the original run with its trigger inputs). - Report the result: Succeeded/Failed with action details.
- Offer to call
Output Format
Diagnosis Summary
- Flow: [name] ([ID])
- Run: [run ID] | [start time] | Status: [status]
Failed Actions
| # | Action | Status | Error Code | Error Message |
|---|
Root Cause
Fix
[Step-by-step fix with code]