Power automate run forensics
Skill Wals-pro/claude-power-automate/skills/power-automate-run-forensics
A Claude Code suite for Microsoft Power Automate development — Agent Skills + a local CLI (Dataverse/Power Platform APIs, Azure CLI auth, no stored secrets). Beta.
npx -y skills add Wals-pro/claude-power-automate --skill power-automate-run-forensicsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 8 stars8 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Diagnose Microsoft Power Automate cloud-flow incidents from run history, failed actions, trigger behavior, connection failures, missing runs, skipped branches, invalid expressions, throttling, retries, and action outputs before changing workflow code. Use with the power-automate CLI runs and run-detail commands for read-only evidence gathering.
SKILL.md
1.5 KB, as published. Nobody here has run it
Power Automate Run Forensics
Start read-only. Confirm what the flow actually did before changing JSON, connections, or production records.
Procedure
- Define boundary: profile, environment ID, flow ID/name, run ID(s), object IDs, and absolute time window.
- Run
power-automate status --profile <profile>andpower-automate runs --profile <profile>when API access is available. - If the flow or environment is unknown, run
power-automate environments,power-automate flows, orpower-automate runs --all-environments --top <n>to discover what the signed-in Azure user can see. - For a failed or suspicious run, inspect
power-automate run-detail. - Compare trigger data, action inputs, action outputs, branch status, skipped actions, and connector error codes.
- Read
references/errors-limits.mdfor Microsoft error/limit guidance. - Produce findings before proposing edits.
Output
Report:
- Boundary checked.
- Evidence reviewed.
- Findings with run IDs and timestamps.
- Impact.
- Root cause or strongest hypothesis.
- Fix target and validation plan.
If evidence is missing, name the missing source and why it matters.