Document business processes with AI
Interview a process owner and preserve the actual workflow in matching Markdown and JSON files for review and a later builder handover.
Document business processes with AI through an interview about how work actually happens. The process-mapper skill asks about scope, stages, systems and judgement calls, then writes matching Markdown and JSON records for someone who may later build an automation.
The interview does not produce a deployed workflow. Exact field labels, evidence and exceptions matter because vague descriptions leave the next builder guessing. Rule-based choices remain separate from decisions that need human judgement, and both output views should represent the same facts.
What it does
- Record actual steps, inputs and outputs
- Distinguish decision rules from judgement calls
- Clarify responsible and approval roles
- Capture exact systems and field labels
- Preserve common failures and recovery paths
- Prepare matching files for a builder handover
Install and use
- Load the complete folder in a file-reading and writing agent and choose an output location. Bring someone who knows the process and the necessary examples; Drive and Slack connections are optional.
- Ask: “Interview me about purchase-request approval. Record actual steps and exceptions, then deliver both files. Send no notifications.”
- Correct the sequence and details after each stage. Provide exact labels or tightly scoped screenshots, removing unrelated customer information.
- Review the readable document and JSON together so changes appear in both. Separately choose any synced destination or completion notice and check the actual handover.
FAQ
Source
Made by Justin Brock. Original skill: process-mapper, published on GitHub under the MIT licence.
View on GitHub