Scope Check

Feature Creep Guard

Before adding anything, answer one question: What are the 2-3 things this project MUST do well? Everything else is secondary.

Evaluation Framework

For any proposed addition, score each dimension:

1. Core Goal Alignment

Does it make the primary use case more compelling, or does it distract?

• Helps - Directly improves the thing users come here for
• Neutral - Doesn't hurt, but doesn't serve the main purpose
• Hurts - Splits focus, confuses the product's identity

2. Complexity Cost

What burden does this add to the codebase and to users' mental models?

• None - Invisible to users and developers
• Minimal - One new concept, well-isolated
• Significant - Touches multiple systems, adds configuration, needs documentation

3. Implementation Cost

How long to build, and what's the opportunity cost?

• Trivial - Hours, no new dependencies or infrastructure
• Moderate - Days, some new patterns or integrations
• Substantial - Weeks, new systems, significant testing surface

4. Architectural Fit

Does it work naturally with what exists?

• Natural - Extends existing patterns, uses existing infrastructure
• Acceptable - Some adaptation needed, but no architectural compromises
• Forced - Requires workarounds, violates existing patterns, introduces tech debt

5. Maintenance Burden

What's the ongoing cost after shipping?

• None - Works indefinitely once shipped
• Low - Occasional updates with dependency bumps
• Ongoing - Content updates, API changes, infrastructure monitoring, support load

Decision Matrix

| Alignment | Complexity | Cost | Fit | Maintenance | Verdict | |-----------|-----------|------|-----|-------------|---------| | Helps | Any | Any | Natural/Acceptable | Any | ADD | | Neutral | None/Minimal | Trivial | Natural | None/Low | ADD (low risk) | | Neutral | Significant | Moderate+ | Any | Ongoing | DEFER | | Hurts | Any | Any | Any | Any | REJECT | | Any | Any | Substantial | Forced | Ongoing | REJECT (too expensive) |

Output Format

FEATURE: [proposed addition]
VERDICT: ADD / DEFER / REJECT

  Core alignment:    [helps / neutral / hurts]
  Complexity cost:   [none / minimal / significant]
  Implementation:    [trivial / moderate / substantial]
  Architectural fit: [natural / acceptable / forced]
  Maintenance:       [none / low / ongoing]

REASONING: [1-2 sentences on the decision]
ALTERNATIVES: [simpler way to achieve the same goal, if any]
REVISIT: [conditions under which to reconsider, if DEFER]

Common Traps - Push Back On These

• "Nice to have" that doesn't serve the core use case - DEFER until core is bulletproof
• Configuration options that could just be good defaults - ship the default, add config if users complain
• Premature abstractions for extensibility that may never be extended - build for today's requirements
• "What if" features driven by hypothetical users rather than actual feedback - wait for real signal
• "Small additions" that are actually scope increases in disguise - evaluate the full cost, including testing and maintenance
• Parity features ("competitor X has it") - unless it serves YOUR core goals, it's distraction
• "While we're at it" additions during refactors - refactoring scope should be strictly mechanical

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: pvnarp
• Source: pvnarp/agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.