Motivation
Level 3 requires the user to make a "substantive modification" to AI code before proceeding, and the agent must reject "style changes" as insufficient. However the current definition is purely qualitative:
Require substantive modifications that demonstrate understanding (not just style changes)
Without concrete criteria, the agent will be inconsistent in what it accepts. Users may also feel unfairly blocked or game the system without a clear bar.
Proposed Change
Add an explicit rubric to the Level 3 section. A modification is substantive if it:
- Changes logic, control flow, or algorithm (not just renaming or reformatting)
- Adds or removes a guard condition or edge-case handler
- Substitutes a different data structure or type
- Refactors a section in a way that changes how it executes
- Adds a meaningful comment that explains a non-obvious decision
A modification is not substantive if it only:
- Renames variables or functions (without changing their meaning)
- Reformats whitespace or indentation
- Reorders lines without changing logic
- Adds "// looks good" or similar cosmetic comments
The agent response when rejecting a superficial change should name the rubric explicitly, e.g.:
That change is cosmetic only (variable rename). A substantive modification
needs to affect logic or structure. For example, you could: [specific suggestion].
Motivation
Level 3 requires the user to make a "substantive modification" to AI code before proceeding, and the agent must reject "style changes" as insufficient. However the current definition is purely qualitative:
Without concrete criteria, the agent will be inconsistent in what it accepts. Users may also feel unfairly blocked or game the system without a clear bar.
Proposed Change
Add an explicit rubric to the Level 3 section. A modification is substantive if it:
A modification is not substantive if it only:
The agent response when rejecting a superficial change should name the rubric explicitly, e.g.: