Development & Code Intelligence Tool

The Code Smell Detector Tool

Run lightweight static heuristics across a repository snapshot to surface the maintainability risks — the long functions, duplication, and complexity — that quietly accumulate into technical debt, on infrastructure you control.

Explore VDF AI Agents
StaticNo execution required
Repo-wideScans the whole snapshot
DebtQuantifies maintainability risk
100%On-prem, code never leaves
The Debt Problem

Technical debt is invisible until it’s expensive

Smells like sprawling functions, duplication, and tangled complexity don’t break the build — they just make every future change slower and riskier. By the time they’re obvious, they’re costly to fix.

01

Debt is invisible

Smells pass every test while quietly raising the cost of change.

02

Perception is subjective

"This code is messy" isn’t actionable without consistent signals.

03

It compounds silently

Each unaddressed smell makes the next change a little harder.

04

Hosted analysis sees code

Quality scanning can’t mean shipping your source to a third party.

How the Tool Works

Consistent smell heuristics

Detection

Heuristics for the usual suspects

Length, duplication, complexity.

The tool applies lightweight static heuristics across a repository snapshot to flag the recurring smells — overly long units, duplication, and complexity — turning a vague sense of mess into concrete, comparable signals.

  • Function/length heuristics
  • Duplication signals
  • Complexity flags
  • Whole-repository coverage
Static
Heuristics

No execution needed

LengthDuplicationComplexitySignals

Prioritization

Where to refactor first

Debt you can rank.

By surfacing smells consistently across the repo, an agent can point teams at the highest-risk areas to refactor first instead of guessing where the debt is worst.

Ranked
Refactor Targets

Highest risk first

PrioritizeHotspotsRefactorPlan

Governance

On-premise analysis

Source stays internal.

Analysis runs against a snapshot inside your perimeter with audit logging, so quality scanning never means exposing your code.

100%
On-Prem

IP-safe, logged

On-premIP-safeAudit logPrivate
Inputs

Parameters

The code_smell_detector tool accepts these inputs when an agent calls it. Required inputs are flagged.

Name Type Required Description
owner string Optional GitHub repository owner.
repo string Optional GitHub repository name.
ref string Optional Branch, tag, or commit SHA. Defaults to the default branch.
user_id integer Optional User ID for resolving a stored GitHub token.
Where it pays back

Where smell detection pays back

Refactor planning

Find the highest-debt areas to prioritize for refactoring.

Quality gates

Flag changes that introduce new smells before merge.

Legacy assessment

Quantify the maintainability of an inherited codebase.

Tech-debt reporting

Give leadership a consistent read on code health.

Onboarding context

Show new engineers where the rough edges are.

Agent workflows

Let an engineering agent track debt across a network.

How VDF AI connects it

Assigned to agents, orchestrated as networks

On VDF AI, an industry’s use cases map to agents, and you assign tools like this one to those agents. Compose multiple agents into a governed, on-premise network.

ROI Snapshot

What changes after you assign it

Visible
Debt made measurable
Prioritized
Refactoring by risk
Consistent
Same lens across the repo
100%
Analyzed on-prem
FAQ

Questions about the Code Smell Detector tool

What is the code smell detector?

It runs lightweight static heuristics over a repository snapshot to surface maintainability smells — long units, duplication, complexity — as concrete signals. Assigned to an agent, it makes technical debt visible and rankable.

How is it different from code review?

Code review evaluates correctness and security on specific code; the smell detector takes a repo-wide, structural view of maintainability so teams know where debt concentrates.

Does it execute the code?

No. It is purely static, applying heuristics to the source in a snapshot, so it is safe and fast to run repeatedly.

Is our code exposed?

No. Analysis runs on-premise with audit logging; your source never leaves your perimeter.

Which agents use it?

Engineering and strategy agents use it alongside the repository map and code review tools to plan and track refactoring.

Make technical debt measurable

See the code smell detector help an engineering agent prioritize refactoring — on-premise.