Agent Skills For Go

AI Agent Skills for writing idiomatic, production-quality Go code. 20 modular skills teach AI coding assistants Go best practices derived from:

• Google Go Style Guide
• Effective Go
• Uber Go Style Guide
• Go Wiki CodeReviewComments

Skills are tuned following agentskills.io best practices: content the agent already knows is omitted, procedural decision trees guide multi-step tasks, 48 reference files load on demand via progressive disclosure, 8 bundled scripts automate common checks, and 4 asset templates ensure consistent output.

Skills Included

Skill	Description
go-code-review	Systematic checklist for reviewing Go code and PR submissions
go-concurrency	Goroutine lifecycle, channels, mutexes, parallelization, thread-safety
go-context	Context.Context placement, cancellation, deadlines, request-scoped data
go-control-flow	Idiomatic conditionals, loops, switch/break behavior, guard clauses
go-data-structures	Slices, maps, arrays — allocation with new vs make, append, copying
go-declarations	Variable/const/type declarations, var vs :=, iota enums, shadowing
go-defensive	API boundary hardening, defer cleanup, Must functions, time handling
go-documentation	Doc comments, package docs, godoc formatting, runnable examples
go-error-handling	Error strategy decisions, wrapping (%v vs %w), sentinels, logging patterns
go-functional-options	Functional options pattern for constructors with optional config
go-functions	Function ordering, signature formatting, Printf verbs, Stringer interface
go-generics	When to use generics, constraints, common pitfalls, type aliases
go-interfaces	Interface design, abstractions, embedding, "accept interfaces return structs"
go-linting	Linters, golangci-lint setup, nolint directives, CI/CD integration
go-logging	Structured logging with slog, log levels, request-scoped context, migration
go-naming	Naming decision flow for packages, types, functions, variables, receivers
go-packages	Package organization, imports, package size, CLI/flag patterns
go-performance	String optimization, capacity hints, benchmarking, strconv over fmt
go-style-core	Formatting, nesting reduction, style principles, fallback style guide
go-testing	Table-driven tests, subtests, test helpers, assertions, test organization

Bundled Scripts

8 scripts automate common Go checks. All support --help, --json for structured output, and meaningful exit codes (0 = clean, 1 = issues found, 2 = error). Analysis scripts support --limit to cap output size, and destructive scripts require --force to overwrite existing files.

Script	Skill	Purpose
pre-review.sh	go-code-review	Run gofmt + go vet + golangci-lint before review
check-naming.sh	go-naming	Detect SCREAMING_SNAKE, Get-prefixed getters, bad package names
check-docs.sh	go-documentation	Find exported symbols missing doc comments
check-errors.sh	go-error-handling	Catch bare returns, string comparison on errors, log-and-return
check-interface-compliance.sh	go-interfaces	Find interfaces missing compile-time verification
bench-compare.sh	go-performance	Run benchmarks with optional benchstat comparison
setup-lint.sh	go-linting	Generate .golangci.yml with recommended linters
gen-table-test.sh	go-testing	Scaffold a table-driven test file

Quick Install

Using npx skills (Recommended)

The easiest way to install across any AI coding agent. Supports Cursor, Codex, OpenCode, Cline, GitHub Copilot, Windsurf, Roo Code, and 25+ more agents.

npx skills add cxuu/golang-skills --all

Claude Code

# Add the marketplace (one time)
/plugin marketplace add cxuu/golang-skills

# Install the skills
/plugin install golang-skills@cxuu-golang-skills

Cursor (Native Remote Rule)

1. Open Cursor Settings (Cmd+Shift+J on Mac, Ctrl+Shift+J on Windows/Linux)
2. Navigate to Rules → Add Rule → Remote Rule (Github)
3. Enter: https://github.com/cxuu/golang-skills

How It Works

These skills follow the Agent Skills open standard, which works across multiple AI coding tools. When you're writing Go code:

1. Automatic activation: The AI agent loads relevant skills based on context (e.g., go-naming when you're writing a new function)
2. Procedural guidance: Decision trees and step-by-step procedures for multi-step tasks like code review and error strategy selection
3. Progressive disclosure: Core rules load immediately; 48 reference files load on demand when specific situations arise
4. Automation: 8 bundled scripts handle repetitive checks so the agent focuses on higher-level guidance
5. Conditional cross-references: Skills link to each other with "when" conditions to avoid unnecessary context loading
6. Rule ownership: docs/RULE_OWNERSHIP.md keeps duplicated guidance out of non-owner skills

Project Structure

.
├── skills/
│   └── go-*/
│       ├── SKILL.md      # Core rules (< 225 lines each)
│       ├── references/   # Detailed guidance, loaded on demand
│       ├── scripts/      # Automation scripts and helpers
│       └── assets/       # Output templates (4 skills)
├── evals/
│   ├── evals.json        # Trigger and quality eval definitions
│   ├── files/            # Sample Go files for quality evals
│   └── fixtures/         # Test fixtures for script/eval coverage
├── docs/                 # Repository maintenance notes
├── .github/workflows/    # CI validation
└── source/               # Original style guide sources

Provenance and Compatibility

Bundled upstream source snapshots live under source/. Each source file keeps its own inline provenance header, and THIRD_PARTY_NOTICES.md summarizes the source path, upstream project, URL, license, and copyright at the repository level.

Go-version-sensitive guidance is tracked in COMPATIBILITY.md. When a skill recommends a standard-library API that depends on a specific Go release, the guidance should name the minimum Go version and include an older fallback where that helps users on maintained but older toolchains.

License

Project-authored skill files, scripts, assets, docs, and evals are licensed under the Apache License, Version 2.0. See LICENSE for details. Bundled upstream snapshots under source/ retain their upstream licenses; see THIRD_PARTY_NOTICES.md.