coding-standards
Canonical cross-language coding standards reference. Shared rules embedded by reviewer agents. Activate when: viewing coding standards, checking naming rules, reviewing style baseline, consulting style guide, what are the rules.
How do I install this agent skill?
npx skills add https://github.com/zereight/gitlab-mcp --skill coding-standardsIs this agent skill safe to install?
- Gen Agent Trust Hubpass
This skill is a static reference for coding standards and best practices. It contains no executable code, network operations, or security risks.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
Coding Standards
D9 Canonical Reference. This is the single source of truth for cross-language coding standards. Language-specific reviewers (
@typescript-reviewer,@python-reviewer, etc.) embed these rules. Agents cite this skill as "See also: /coding-standards."
Naming Conventions
Variables & Functions
| Language | Variables | Functions | Constants |
|---|---|---|---|
| TypeScript / JavaScript | camelCase | camelCase | SCREAMING_SNAKE_CASE |
| Python | snake_case | snake_case | SCREAMING_SNAKE_CASE |
| Go | camelCase | camelCase | CamelCase (exported) |
| Rust | snake_case | snake_case | SCREAMING_SNAKE_CASE |
| Java | camelCase | camelCase | SCREAMING_SNAKE_CASE |
| C# | camelCase | PascalCase | PascalCase |
| Swift | camelCase | camelCase | camelCase |
Classes / Types / Interfaces
PascalCase — all languages, no exceptions.
Booleans
Prefix with is, has, can, should: isLoading, hasError, canEdit, shouldRefresh.
Collections
Plural nouns: users, errors, items — not userList, errorArray.
Avoid
- Single-letter variables outside loop counters (
i,j,kare OK in loops) - Abbreviations that save under 3 characters:
usr→user,mgr→manager - Redundant type names:
UserInterface,UserClass,UserObject→ justUser
Function Design
- Max length: 50 lines (firm guideline; > 80 lines is always a split target)
- Single responsibility: one function, one job — if "and" appears in the description, split it
- Max parameters: 3; beyond that, use an options/config object
- Cyclomatic complexity: ≤ 10; > 15 is a mandatory refactor target
- Nesting depth: ≤ 3 levels; use early returns to flatten
Early Return Pattern (preferred)
// BEFORE — deep nesting
function handle(input) {
if (input) {
if (input.valid) {
return process(input);
}
}
return null;
}
// AFTER — early returns
function handle(input) {
if (!input || !input.valid) return null;
return process(input);
}
Error Handling
- Never swallow errors silently:
catch (e) {}is always wrong - Error messages must contain context:
"Failed to fetch user id=42"not"Error" - Propagate or handle: either handle the error at the right level OR re-throw it — never both and never neither
- Typed errors (TypeScript):
class NotFoundError extends Error { constructor(id: string) ... }not genericnew Error - Python exceptions: catch specific exception types; bare
except:is forbidden - Go errors: always check returned errors; use
errors.Is()/errors.As()for comparison - Rust results: use
?for propagation; no.unwrap()in library code
Code Structure
Immutability-First
constoverlet(JS/TS);valovervar(Swift/Kotlin);finalwhere appropriate- Mark fields
readonlywhen not reassigned after construction - Prefer immutable data structures for function arguments
No Magic Numbers
// BAD
if (retries > 3) { ... }
setTimeout(fn, 5000);
// GOOD
const MAX_RETRIES = 3;
const POLL_INTERVAL_MS = 5000;
if (retries > MAX_RETRIES) { ... }
setTimeout(fn, POLL_INTERVAL_MS);
No Commented-Out Code
If it is dead → delete it (git history preserves it). If it is needed soon → it should be in a branch. If it explains a non-obvious decision → keep it as a comment, not commented-out code.
Anti-Patterns Reference
| Pattern | Severity | Reason |
|---|---|---|
| Mutable global state | HIGH | Unpredictable side effects; hides dependencies |
| Promise not awaited | HIGH | Unhandled async errors silently swallowed |
any in TypeScript | MEDIUM | Bypasses type safety across call boundaries |
console.log in production code | LOW | Log noise; potential data leak in sensitive contexts |
TODO without issue tracker reference | LOW | Becomes permanent tech debt |
| God Object | HIGH | Single class with too many responsibilities |
| Magic numbers inline | MEDIUM | Unclear intent; maintenance hazard |
| Copy-paste logic | MEDIUM | Silent divergence over time |
| Catching and re-throwing without context | MEDIUM | Stack traces lose meaning |
| Nested ternary operators | MEDIUM | Unreadable; use if/else or switch instead |
SOLID Principles Checklist
- Single Responsibility: does this class/function do exactly one thing?
- Open/Closed: extend via composition/interfaces, not inheritance modification?
- Liskov Substitution: can a subtype always replace the base type without breaking callers?
- Interface Segregation: no fat interfaces — clients should not depend on methods they don't use?
- Dependency Inversion: depend on abstractions (interfaces), not concretions?
See Also
@code-reviewer— applies these rules during code review@typescript-reviewer,@python-reviewer, etc. — language-specific rules with these as baseline@security-reviewer— security-specific standards (OWASP, secrets, crypto)
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/zereight/gitlab-mcp/coding-standards">View coding-standards on skillZs</a>