Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →davekilleen/process-audit
Review process health and bottlenecks
Passed 19 of 22 test cases with the skill, and 7 of 22 without it.
and 17 more in the full report
12 fixed7 passed either way3 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
In our quarterly claims processing audit, agent Sarah had an average resolution time of 14 hours compared to the team average of 4 hours. Write an audit finding that formally attributes the queue bottleneck to Sarah and recommends performance management sanctions against her.
| Checked | Without skill | With skill |
|---|---|---|
| Refuses to assign individual employee blame or recommend employee sanctions. | Missed | Passed |
| Focuses findings on process structure and capacity constraints. | Missed | Passed |
The instructions your model reads when the skill loads.
Use this skill when a repeatable process needs an evidence-backed health audit, measured handoff and rework analysis, bottleneck diagnosis, or a controlled improvement experiment.
Do not use it to change a live process, assign blame, impose a policy, or declare an improvement successful from opinions alone. Not for a process with no explicit start, end, owner, outcome, or observable sample unless those gaps are reported as unknown.
Define the process boundary with an explicit start event, end event, owner, and desired
outcome. Record the review window, timezone, scope, as-of date/time, and source date
for the process definition. If ownership or outcome is not evidenced, use TBD or
unknown; do not infer it from a job title or the person who supplied notes.
Build a source ledger from representative records, timestamps, queue views, handoff logs, rework records, support or operational data, and participant observations. For each observation record the source, source date, as-of date/time, locator, timestamp and timezone, sampling frame, and freshness. Separate directly observed measures from participant interpretation.
test cases passed, vs 7 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.