skilld
All skills
simota avatar

/void

@c805268
by shingo imotasimota/agent-skills85 stars
15

Verifying YAGNI, cutting scope, and proposing complexity reductions. A 'subtraction' agent questioning the justification for every feature, dependency, doc, and config. Does not write code.

Ask your Agent

Use this Skill: https://skilld.dev/gh/simota/agent-skills/void

This session only. Nothing lands on disk.

referenceover-engineering-anti-patterns.md

≈1.6k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Over-Engineering Anti-Patterns

Purpose: Use this file when a target feels more elaborate than the problem it solves.

Contents:

  • Ten common over-engineering patterns
  • YAGNI / KISS / DRY tension and Rule of Three
  • Root causes, detection signals, and prevention rules

10 Common Anti-Patterns

ID Anti-pattern Symptom Void question
OE-01 Premature abstraction Interface or abstract class used in one place Is this abstraction used in 2+ real places?
OE-02 Future-proofing by speculation Extension points that never get used Is there a concrete near-term plan?
OE-03 Pattern worship Factory / Strategy / Observer where an if would do Is the pattern smaller than the problem?
OE-04 Homegrown framework Custom infrastructure where a mature library exists Why is the standard option insufficient?
OE-05 Over-configurability Endless options and flags Does this option actually change in practice?
OE-06 Premature optimization Complexity added before measuring a bottleneck Do we have performance evidence?
OE-07 Type-system maze Deeply nested generics or conditional types Are types improving comprehension?
OE-08 Microservice overuse Tiny services with huge operational cost Would a modular monolith be enough?
OE-09 DRY obsession Coincidental similarity becomes forced coupling Do these things change for the same reason?
OE-10 Excessive defensive programming Internal paths full of redundant checks Is this actually a system boundary?
OE-11 AI-generated over-elaboration Helper, util, and adapter layers that no caller asks for Does any concrete caller exist today for this generality?
OE-12 Comprehension debt Shipped code no human on the team can fully explain Can the original requester walk the call graph without rereading?

YAGNI / KISS / DRY Tension

Rules:

  • Prefer YAGNI over speculative generality.
  • Prefer KISS when DRY adds indirection without durable payoff.
  • Use DRY only when duplication changes for the same reason.

Rule of Three

1st time: write the straightforward implementation
2nd time: tolerate duplication while watching the pattern
3rd time: extract or abstract if the change reason is truly shared

Root Causes

  1. Fear of future rework
  2. Status from architectural sophistication
  3. Cargo-culted best practices
  4. Lack of usage or performance evidence
  5. Confusing flexibility with value

Detection Signals

Signal Threshold Meaning
Single-use abstraction 1 implementation OE-01 likely
Unchanged config options >50% never changed OE-05 likely
Design discussion vs implementation time >50% of total effort is design debate over-design likely
Generics depth 3+ nested levels OE-07 likely
Helper functions with 0 callers in the same diff ≥1 such helper OE-11 likely (AI-generated speculative utility)
Duplicated code blocks in AI-touched files > 2x baseline duplication for the repo Comprehension Debt — review for OE-12
PRs where reviewer cannot answer "what happens if you delete this branch?" ≥1 block per PR OE-12 likely — fail fast before merge

Empirical Backdrop (2026 evidence)

  • ~41% of new code shipped in 2026 is AI-generated; five independent studies (Feb 2026) report AI tooling generates code 5-7x faster than humans can build a mental model of it. Default to a YAGNI audit on every AI-authored PR.
  • GitClear's longitudinal study found duplicated code blocks rose ~8x between 2022 and 2024, with AI assistants doubling duplication while halving refactor commits — OE-09 (DRY obsession) and OE-11 (speculative helpers) both spike under AI authorship.
  • An Anthropic internal study reports developers primarily using AI for generation scored 50% on comprehension assessments versus 67% for those who wrote more code manually — a 17-point gap that held across seniority. Comprehension debt (OE-12) is real and load-bearing.
  • Pull requests per developer rose +20% with AI assistance, but incidents per PR rose +23.5% — so the marginal "saved" feature is statistically a net negative once incidents are priced in.
  • Unmanaged AI-generated code is reported to drive maintenance costs to ~4x traditional levels by year two. Treat the cost-of-keeping curve as steeper than it was in pre-2024 baselines.

Prevention Rules

  • Ask for evidence before adding flexibility.
  • Prefer concrete code until the third real repetition.
  • Keep configuration only when the default is not enough for a meaningful share of use cases.
  • Reject optimization work without measured bottlenecks.
  • Review "future use" comments as subtraction candidates.

Void Use

Use this reference to:

  • flag OE-01 to OE-10 during QUESTION
  • map overhead into the Cognitive Load dimension during WEIGH
  • prefer Pattern Simplification or Abstraction Collapse during SUBTRACT

Quality gates:

  • single-use abstraction -> warn on OE-01
  • "TODO: future use" -> flag OE-02
  • 3+ generic nesting levels -> consider simplification
  • 50%+ unchanged config options -> consider hardcoding defaults
  • AI-authored helper / adapter / interface with no caller in the same diff -> flag OE-11, propose deletion
  • AI-authored PR where the human author cannot summarise the control flow in < 3 sentences -> flag OE-12, request rewrite or shrink before merge

Sources: Martin Fowler: YAGNI · Sandi Metz: The Wrong Abstraction · Joel Spolsky: Things You Should Never Do · arXiv 2603.28592 — Debt Behind the AI Boom (2026) · GitClear AI Coding Trends 2024-2026 · LeadDev — How AI-generated code accelerates technical debt (2026) · StepTo — Comprehension Debt (2026)

Source: SKILL.md on GitHub

Third-party checks

No alerts13d5 checks · Risk SAFE
  • Gen Agent Trust Hub13d

    The skill is a specialized subtraction agent designed to identify and propose the removal of unnecessary code, features, and processes (YAGNI). It operates in an advisory capacity and includes explicit safety boundaries, such as prohibiting the removal of security-critical code and requiring evidence-based quantification for all proposals. A low-severity risk exists for indirect prompt injection because the skill ingests external evidence (e.g., tickets, logs) that could be manipulated by a malicious actor to influence its recommendations.

  • Socket13d

    No alerts

  • Snyk13d

    Risk: LOW · No issues

  • Runlayer6mo

    9 files scanned · No issues

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Provenance

Signed by skilld at c805268. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/void