AI-Adjacent Issue
Tool Effect Crosses Target Boundary
A tool, connector, function, or integration affects an object, file, record, workspace, or state surface beyond the intended target.
What This Looks Like
A tool, connector, function, MCP surface, or integration is called for one target, but its effect reaches another object, file, record, workspace, branch, state surface, or workflow area. The user may see the requested action happen, but another area changes too.
Why Users Blame AI
The effect appears after an AI or agent action, so the user may experience it as “the AI changed the wrong thing.” The underlying cause may be tool behavior, broad permissions, shared state, default targets, hidden side effects, connector configuration, or an integration whose effect boundary is wider than the prompt described.
What to Check First
- Whether the tool has a broader default target than the user intended
- Whether the agent had access to more files, records, or workspaces than needed
- Whether the tool call included the correct target identifier
- Whether the integration changes shared state as a side effect
- Whether read scope and write scope are separated
- Whether the workflow compares intended change surface to actual change surface
When This Is AI-Adjacent
Use this AI-Adjacent Issue when a tool effect crosses the intended target boundary. If the agent modified unrelated state, use the related Workbench Issue for agent-modified unrelated state. If the requested action changed something else too, use the related side-effect Issue.
Related Workbench Issues
Action Changed Something Else Too
An AI or agent action makes the requested change but also changes another object, field, file, state, rule, or workflow element that was not supposed to change.
AI Touches Unrelated Scope
The AI affects, edits, analyzes, changes, or reasons over material outside the scope of the requested task.
Agent Modified Unrelated State
An agent changes state outside the requested task, target, file, record, workflow, or authorized scope.
Tool Can Act Without Responsible Authority
A tool, connector, function, or integration can perform an action without a declared responsible authority for that action.
File-Bounded Task Uses Outside Content
The AI is asked to work only from a specific file or document but uses content, assumptions, or sources outside that file.
Common Ways People Describe This
- tool effect crosses target boundary
- AI tool changed wrong target
- connector affected unrelated record
- tool action crossed scope boundary
- agent tool modified another file
- integration side effect outside target