PENGO Workbench
Workbench / Issues / Small Change Produces Large Downstream Effects

Issue

Small Change Produces Large Downstream Effects

A small prompt, schema, policy, output, or workflow change creates unexpectedly large effects in downstream steps.

What This Looks Like

A small change to a prompt, field, schema, policy, example, model setting, tool contract, or workflow step produces a much larger effect later in the system. The immediate change may look minor, but downstream outputs, routes, parsers, reviews, or decisions shift more than expected.

Why It Matters

Small changes become risky when their downstream effects are not visible. A user may approve a minor edit without realizing it changes later behavior, breaks compatibility, or alters review outcomes. This makes regression testing and change control harder.

Structural Signal

A local change propagates into downstream effects beyond its apparent size or declared scope. The issue is not that change occurred; it is that the change has an amplified effect through connected workflow structure.

Common Triggers

  • A small prompt edit changes output structure used by later steps
  • A field label changes while parsers still expect the old label
  • A policy wording change affects routing or escalation
  • Examples are updated and shift model behavior broadly
  • A tool schema changes in a way that breaks dependent prompts
  • The workflow lacks dependency mapping for small changes

When to Use This Issue

Use this Issue when a small change creates large downstream effects across AI output, workflow behavior, tooling, review, routing, or structured data.

When Not to Use This Issue

Do not use this Issue when a large downstream effect was expected and declared. Do not use it for ordinary local edits that remain contained.

Category

Spread & Escalation

Primary Pattern

PAT-0360 — Propagation Amplification

Declared Patterns

Derived Primary Lenses

Derived Secondary Lenses

Related AI-Adjacent Issues

Search Intents

  • small change produces large downstream effects
  • small AI change caused big workflow impact
  • prompt change had large downstream effect
  • minor schema change broke workflow
  • small update caused major AI behavior change
  • downstream effects from small change