AI Automation Change Impact Pack

Compare one sanitized agreed automation baseline with one new change request. The pack deterministically traces direct targets and transitive dependents across triggers, integrations, data, permissions, exception/retry behavior, proof, acceptance, ownership, and support.

What you get

• A private change-impact report • A test-and-acceptance delta • A machine-readable JSON record • A separate client-safe decision page

Contradictions, missing evidence, unknowns, limitations, and support boundaries stay visible. The final approve, clarify, defer, or decline decision remains with a human.

Use cases

• A client asks to replace an integration after the baseline is agreed. • A field-map or permission change may affect downstream retry and acceptance paths. • An agency needs private technical analysis and a separate client-safe decision page without drafting a legal change order.

Known limitations

• Supplied sanitized evidence only; no live automation is inspected or executed. • Structured JSON input is required. • Optional exports, schedules, estimates, and test summaries remain untrusted. • No time, price, approval, test result, owner, or support promise is invented. • No legal drafting, compliance assurance, automatic client sending, or marketplace/production mutation. • Credentials, secrets, full customer datasets, regulated records, and unnecessary personal data are refused.

Sample input and output

Use examples/golden-input.json from the package. It is illustrative sanitized data, not reusable client evidence. The exact run creates private-change-impact.md, test-acceptance-delta.md, impact-record.json, and client-decision.md.