skills.rest
ooooooooooooooooooop
ooooooooooooooooooopCOMMUNITY
Explore all skills by this author →
GitHub

minimal-implementation

Constrains code changes to the minimal correct scope with verifiable evidence.

3|Updated Feb 13, 2026
Search Tags:#code review#refactoring#bug fix#verification#minimal change#scope control#code modification
One-click install
npx skills add https://github.com/ooooooooooooooooooop/personal-ai --skill minimal-implementation-ooooooooooooooooooop
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: minimal-implementation
Source: https://github.com/ooooooooooooooooooop/personal-ai/tree/main/skills/minimal-implementation
Command: npx skills add https://github.com/ooooooooooooooooooop/personal-ai --skill minimal-implementation-ooooooooooooooooooop

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI agents often over-engineer tasks by adding unnecessary abstractions, fallback paths, dependencies, and unrelated refactors. This Skill enforces a minimal-change discipline so modifications stay small, correct, and easy to review. ## Core Features & Use Cases - Decision Ladder: Before adding any code, rule, helper, or abstraction, it checks whether existing implementations, standard libraries, platform capabilities, or installed dependencies already suffice. - Change Budget & Safety Boundaries: Requires declaring the minimal impact surface, acceptance checks, and files intentionally left untouched; forbids swallowing exceptions, silent degradation, and masking root causes. - Output Contract: Reports the modified surface, reused implementations, verification evidence, deliberately unchanged scope, and residual risks. - Use Case: When asked to fix a parser bug, the agent changes only the loop boundary, adds one regression test, and reports exact test results instead of rewriting the parser. ## Quick Start Use the minimal-implementation skill to fix this bug with the smallest correct change and show me the verification evidence.

Frequently Asked Questions about minimal-implementation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent an AI agent from over-engineering a simple code change?▼

Apply a minimal-change discipline that requires the agent to justify every new abstraction against a decision ladder: check existing implementations, standard library, platform capabilities, and installed dependencies first. The agent stops at the first level that satisfies the request.

How to make an AI report verification evidence for code changes?▼

Require an output contract where the agent reports the modified surface, reused implementations, acceptance evidence, and residual risks. Fresh targeted tests, validators, hashes, or direct behavior checks count as evidence, not exit codes alone.

When should a minimal implementation approach not be used?▼

Expand scope only when the user explicitly requests architecture work, long-term extensibility, or comprehensive refactoring. A generic request to optimize or improve does not automatically justify a full rewrite.

Why does the skill forbid fallbacks and silent exception handling?▼

Unnecessary fallbacks, swallowed exceptions, and silent degradation mask root causes instead of fixing them. The skill requires fixing root causes on shared paths rather than patching visible symptoms or altering test expectations.

Can this approach handle large refactoring tasks?▼

Yes, but only when the user explicitly asks for broad refactoring. For medium and large tasks it still requires a declared change budget, acceptance checks, and a report of intentionally unmodified scope to keep the work reviewable.

Related Skills

Explore More Software Engineering Skills→
M4n5terM4n5ter
3

ask-questions-if-underspecified

Ask targeted clarifying questions before acting on underspecified requests.

Community
Basic
jamescjamesc

resolve-merge

Merge origin/main into the current branch and resolve conflicts.

Community
Advanced
SirTheOracleSirTheOracle

proposal-reviewer

Audit final plans against the codebase for correctness and regressions.

Official
Advanced
uuczuucz
66

moyu-en

Restrict code edits to explicitly requested scope and files.

Community
Advanced