What can CodeScene Single-User MCP do?

CodeScene Single-User MCP focuses on local Code Health workflows. It gives your AI assistant the tools it needs to review code, safeguard AI-generated changes, support refactoring, and explain why Code Health matters. A Single-User access token enables local Code Health analysis tools only, while project-level and API-dependent features remain unavailable.

Core capabilities

1. Code Health score

Use code_health_score to get a quick maintainability score for a file or code selection.

This is useful when you want to:

  • understand how healthy a file is

  • rank candidates for refactoring

  • confirm whether a change improved or reduced maintainability

Code Health 10.0 is the target standard for AI-friendly code.

2. Code Health review

Use code_health_review when you want a fuller explanation of what is lowering Code Health.

This helps your AI assistant:

  • explain the current Code Health score

  • highlight the main code smells and design issues

  • suggest safer next steps before making more changes

3. Pre-commit safeguard

Use pre_commit_code_health_safeguard before committing AI-generated or AI-modified code.

This is the safeguard for uncommitted or staged files and should be run before each commit.

4. Change set analysis

Use analyze_change_set as a broader safeguard before opening a pull request.

This checks the full change set against the base branch and helps catch Code Health regressions before review.

5. Refactoring business case

Use code_health_refactoring_business_case when you want to explain the value of refactoring in business terms.

This helps connect Code Health improvements to outcomes such as:

  • higher development velocity

  • fewer defects

  • lower maintenance cost

6. Built-in explanation tools

CodeScene MCP also includes built-in explanation tools:

  • explain_code_health

  • explain_code_health_productivity

These are useful when you want your AI assistant to explain what Code Health means, why it matters, and how healthier code affects delivery speed, defect risk, and maintainability.

7. AI-powered refactoring with ACE

If you also configure CS_ACE_ACCESS_TOKEN, the code_health_auto_refactor tool becomes available.

ACE is optional and requires an additional license. It currently supports:

  • JavaScript / TypeScript

  • Java

  • C#

  • C++

ACE currently supports these code smells:

  • Complex Conditional

  • Bumpy Road Ahead

  • Complex Method

  • Deep, Nested Complexity

  • Large Method

8. Built-in configuration tools

CodeScene MCP includes set_config and get_config.

These let your AI assistant help configure the server, save settings persistently, and show your current configuration when needed.

What is not included in the Single-User edition

The Single-User edition does not include project-level and API-dependent features such as:

  • project selection

  • technical debt hotspots

  • technical debt goals

  • code ownership lookups

Recommended workflows

Safeguard AI-generated code

Recommended flow:

  1. Run pre_commit_code_health_safeguard on staged or uncommitted files.

  2. If needed, run code_health_review for more detail.

  3. Refactor until Code Health is restored.

  4. Run analyze_change_set before opening a pull request.

Improve unhealthy code

Recommended flow:

  1. Run code_health_review.

  2. Focus on the functions with the most severe issues.

  3. Refactor in small, reviewable steps.

  4. Re-run code_health_review or code_health_score after each significant change.

  5. Use code_health_auto_refactor if ACE is available and the language and code smell are supported.