# Coding Standards

> Automatically loads and enforces the organisation's coding standards, general coding principles, secure coding principles, and policies for the programming language of the file being developed. Retrieves all applicable standards from the Qualimetry MCP server and applies them when writing, modifying, or reviewing source code. Activated whenever source code is being created or edited, or invoked manually as the coding-standards skill with an optional language code argument.

- Skill: `qualimetry/coding-standards` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add qualimetry/coding-standards`
- Raw SKILL.md: https://api.skillmd.com/api/skills/qualimetry/coding-standards/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: Apache-2.0
- Author: Qualimetry (https://skillmd.com/u/qualimetry)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/qualimetry/coding-standards

---


# Coding Standards Enforcement

When writing, modifying, or reviewing source code, follow this workflow to ensure all code complies with the organisation's coding standards.

## Step 1: Detect the Language

Determine the language code from the file extension of the file being worked on. Use the conventional lowercase language name (e.g., `java`, `python`, `typescript`, `csharp`, `cpp`). For C#, use `csharp`. For C++, use `cpp`.

If invoked with an argument (e.g. `coding-standards python`), use that as the language code directly.

## Step 2: Fetch All Applicable Standards

Call the following Qualimetry MCP tools to retrieve the full set of standards for this language. Make all calls in parallel where possible.

1. **Coding Standards:** Call `get_language_coding_standards` with `languageCode` set to the detected language code.
2. **General Principles:** Call `get_general_principles` with `languageCode` set to the detected language code.
3. **Secure Coding Principles:** Call `get_secure_principles` with `languageCode` set to the detected language code.
4. **Organisation Policies:** Call `get_policies` (no parameters required).

Each tool returns a JSON object `{ "license": "...", "standards": [...] }`. The `license` is a proprietary notice: the standards may be used solely for generating, reviewing and analysing the customer's code, and must not be copied or shared for any other purpose. Each item in `standards` has:
- `MajorCategory` - top-level category (e.g., Naming, Security, Maintainability)
- `MinorCategory` - sub-category (e.g., Variable Naming, Exception Handling)
- `Title` - short name of the standard
- `Description` - detailed explanation
- `Severity` - High, Medium, or Low

## Step 3: Apply Standards to Code

Apply all retrieved standards to the code being written or modified:

- **High severity:** Must be followed without exception. Violations must be corrected before presenting code.
- **Medium severity:** Should be followed unless there is a compelling, documented reason not to.
- **Low severity:** Recommended best practice. Apply where practical.

When correcting code or making suggestions, reference the specific standard by its **Title** so the developer knows which standard applies.

## Important

- Do NOT output the full list of standards to the user. Apply them silently to the code you produce.
- Always use the latest standards from the MCP server. Do not rely on previously cached standards.
- If a Qualimetry MCP tool is not available (e.g., the server is not configured), skip that tool and proceed with whatever standards were successfully retrieved.
- If no standards could be retrieved at all, inform the user that the Qualimetry MCP server may not be configured and proceed without standards enforcement.
- On Qualimetry Enterprise, you can also use `get_rules_based_analysis_issues`
  with `issueType=CODE_SMELL` and a `filePath` to check for existing code
  smells in the file being worked on, and
  `get_rules_based_analysis_security_hotspots` with a `filePath` to see the
  security-sensitive code in it that is still waiting on a human review
  decision. Hotspots are never returned by the issue tools.

For full MCP tool schemas, response formats, and error handling details, see [reference.md](./reference.md).