The process in someone's head, measured and written down

Paste the process as one step per line — what happens, who does it, how long it takes, how long the work waits before it, how much of it arrives usable. Your browser measures it for free: lead time against touch time, the activity ratio, the rolled first-pass yield, every handoff, the bottleneck, a swimlane. Then one lane finds the waste and proposes a leaner future state your browser prices, and the other writes the SOP with its RACI.

Each example ships with a saved model run, so you can see a whole diagnosis and a whole SOP — reconciliation included — without signing in and without spending a credit.

Process name

Items per week — optional

Working day · week

How it works today — who starts it, what triggers it, the exceptions, what goes wrong, what you want from it nothing written yet

…or drop a CSV, TSV or text export here Numbered prose works too: 3. Finance checks the budget line @Finance (20 min, wait 1 day, 95%)

The process, measured — free, in your browser, no model

Swimlane

#	step	owner	kind	PT	wait	LT	%C&A	share of LT	load

Roles

role	steps	touch time	share

Waste candidates — the engine's, for the model to judge

category	steps	evidence

Free — no account, no run. Value-stream-mapping definitions, applied as written.

Paste at least two steps with times to price the run.

lane

lead time

activity ratio

rolled %C&A

figures checked

disagreements

The current state, read

The waste, by category

Category	Steps	Finding	Evidence

The future state, applied and priced by your browser

The model proposed these changes as operations; the engine applied them to your table. Rejected changes are shown and counted as disagreements.

#	Change	Why	Applied

Future-state swimlane

#	step	owner	kind	PT	wait	LT	%C&A	from

In words

Impact, in the skill's four measures

Measure	Direction	Engine figure	Note

Implementation plan

Process document

Purpose

Scope

Included

Excluded

RACI matrix — linted in the browser

#	Step	Responsible	Accountable	Consulted	Informed

Process flow

Detailed steps

Exceptions and edge cases

Scenario	What to do

Metrics

Metric	Target	How to measure

Reconciliation with the engine

Every number the model wrote was re-read against the figures the browser computed, the verdict against the engine's posture, and the structure against the table. A disagreement here means the prose, not the arithmetic, is wrong.

Notes on the input

Risks

Next steps

Summary

Raw model reply

The reply did not parse as the structured result, so it is shown as it arrived.

What this does, and what it does not

The measurements are the value-stream-mapping definitions, applied as written: a step's lead time is the wait before it plus its process time; total lead time and total process time are sums along the timeline, a parallel group counting once at its longest member; the activity ratio is total process time over total lead time; the rolled percent complete and accurate is the product of every step's. With a weekly demand, Little's law gives the items in process (demand times lead time) and each step's load against its people. The posture bands — flowing, sticky, stalled — are this desk's own: stalled under a 5% activity ratio or a 50% rolled yield, sticky under 25% or 80%, or with more handoffs than half the steps, or a step loaded past its capacity. Working days and weeks use the hours and days you set. The build compared every figure against an independent Python implementation written from the definitions on 2,000 random processes and 29,646 checks, and confirmed that three wrong definitions do disagree: an activity ratio that ignores waiting, a rolled yield taken as a mean, and a parallel group summed instead of taken at its longest.

The model never computes: it reads these facts, names the waste and proposes changes as operations your browser applies and prices, or writes the SOP whose RACI your browser lints. Every figure it writes is read back against the engine and the disagreements are shown. It does not know your organisation, so what it says about roles and exceptions is only as good as what you paste. Derived from the agent skills @anthropics/process-optimization and @anthropics/process-doc (anthropics/knowledge-work-plugins, operations; Apache-2.0, licence and changes); not affiliated with the skills' authors.

Nothing to hand? Load the — seven steps, three approvals, 2.8% activity ratio — the that bounces between an agent and a team lead, or the written up as an SOP. All three replay a saved run for free.