LinaPro's AI skill library consists of two categories:

• Externally installed skills: Installed globally via npm/npx and available across all projects — OpenSpec, goframe-v2, and find-skills.
• Project-bundled skills: Shipped with the project source under .agents/skills/, loaded automatically when you work inside the project directory with Claude Code or another compatible AI tool.

Skills Overview

Skill	Use case	Prerequisites
openspec	Optional spec-driven workflow tool — recommended for AI dev workflows	Recommended
goframe-v2	GoFrame-specific AI skill for higher-quality backend code	Recommended
find-skills	AI skill marketplace search to help select the right skills	Recommended
frontend-design	Create distinctive, high-quality frontend interfaces	None
frontend-patterns	React / frontend best practices and design patterns	None
git-commit-push	Generate a conventional commit message and push	None
git-worktree	Create an isolated Git worktree for a task	None
karpathy-guidelines	Behavioral guardrails to reduce AI coding mistakes	None
lina-archive-consolidate	Consolidate archived OpenSpec iterations	None
lina-auto-archive	Automatically scan and archive completed OpenSpec changes	OpenSpec CLI required
lina-e2e	E2E test case naming conventions and management	Playwright required
lina-feedback	Track and fix issues reported after an OpenSpec change	None
lina-perf-audit	Comprehensive backend API performance audit (manual only)	Running LinaPro environment
lina-review	Structured code and spec review	None
openspec-explore	Thinking-partner mode for exploring ideas and problems	None
openspec-propose	Generate a complete OpenSpec change proposal in one step	None
openspec-apply-change	Implement changes per an OpenSpec task list	OpenSpec CLI recommended
openspec-archive-change	Archive completed OpenSpec changes	OpenSpec CLI recommended
playwright-cli	Browser automation and Playwright test authoring	Playwright required

Note

The lina-upgrade skill referenced in source and ops documentation handles framework and source-plugin upgrade governance. The repository prompts you to invoke it via your AI tool when an installed source plugin has a pending upgrade, but this skill is not bundled in the .agents/skills/ project directory — consult your AI tool's own skill configuration.

How to Use Skills

Externally installed skills: Install them globally using the commands in each skill's section. Once installed, Claude Code automatically recognizes and applies them in the relevant contexts.

Project-bundled skills: Loaded automatically when an AI tool runs inside the project directory — no setup needed. Invoke a skill by describing your scenario in the conversation, or use a slash command (e.g. /git-commit-push) to trigger it explicitly.

Externally Installed Skills

These three skills must be installed manually. Once installed, they take effect globally across all your projects.

openspec

OpenSpec is the recommended command-line tool for LinaPro's spec-driven development workflow. Once installed, workflow skills such as /opsx:explore, /opsx:propose, /opsx:apply, and /opsx:archive will automatically use it as their underlying engine. It is recommended for the LinaPro AI dev workflow.

When to use: Running any step of the OpenSpec workflow — exploration, proposals, implementation, and archiving.

Installation:

npm install -g @fission-ai/openspec@latest

Verify the installation with opsx --version.

goframe-v2

goframe-v2 is a Claude Code skill purpose-built for LinaPro backend Go code. It embeds GoFrame coding conventions, ORM usage patterns, and best-practice examples. The skill activates automatically when you write or modify backend Go files, implement service interfaces, create APIs, or work with data access layers — improving code generation quality and keeping output aligned with framework conventions.

When to use: Developing lina-core backend code, implementing business interfaces, or writing the data access layer.

Installation:

npx skills add github.com/gogf/skills -g

find-skills

find-skills is an AI skill marketplace search tool that helps you quickly discover and evaluate skills suited to your project. When you need to extend your AI toolchain and aren't sure what already exists, find-skills makes the selection process much faster.

When to use: Looking for new AI skills to bring into the project or evaluating the available skill ecosystem.

Installation:

npx skills add vercel-labs/skills --skill find-skills -g

Project Skills

The following skills are maintained in .agents/skills/ alongside the project source. No installation is required — they load automatically when you use an AI tool inside the project directory.

frontend-design

Guides the AI toward creating distinctive, production-grade frontend interfaces rather than generic "AI slop" aesthetics. Covers design intent analysis (page purpose, tone, differentiation), typography and color palettes, motion design, spatial composition, and background / visual detail treatment.

When to use: Building a new page, redesigning an existing interface, or whenever you want frontend work to have a considered visual direction.

How to invoke: Describe what you want to build and the AI applies these guidelines automatically, or use /frontend-design.

frontend-patterns

Provides the AI with a comprehensive reference for modern frontend development: component composition, compound components, render props, custom hooks, Context + Reducer state management, memoization, code splitting, virtualization, form validation, error boundaries, Framer Motion animations, and keyboard accessibility patterns.

When to use: Implementing frontend components, choosing a state management approach, optimizing performance, or adding accessibility support.

How to invoke: Activates automatically in frontend development contexts, or use /frontend-patterns.

git-commit-push

Reviews the current Git workspace, generates a conventional commit message from the actual diff following the repository's commit format (with type prefixes like fix, feat, build, docs), stages all changes, commits, and pushes to origin.

When to use: After finishing a feature or fix and ready to commit and push.

How to invoke: Say "commit and push" or use /git-commit-push.

git-worktree

Creates a dedicated Git worktree so the AI continues work in a fresh, isolated directory without touching the current checkout. Especially useful when juggling multiple feature branches in parallel.

When to use: Isolating a development task from the main working tree, or working on multiple branches simultaneously.

How to invoke: Use /git-worktree — the AI derives a branch name from the task description automatically.

karpathy-guidelines

A set of behavioral guardrails that reduce common LLM coding mistakes, inspired by Andrej Karpathy's observations on LLM coding behavior.

Four core principles: Think before coding (state assumptions and tradeoffs explicitly), Simplicity first (write the minimal code that solves the problem), Surgical changes (only touch what is necessary), Goal-driven execution (define verifiable success criteria and loop until verified).

When to use: All coding tasks — these guidelines act as a baseline behavioral constraint for AI-assisted development.

How to invoke: Applied automatically as a baseline; rarely needs explicit triggering.

lina-archive-consolidate

Reads archived OpenSpec iterations from openspec/changes/archive/, groups them by functional responsibility, and performs semantic merging — combining multiple iterations into a single cohesive archive while preserving all design decisions and implementation details. When the archive count reaches 8 or more, it automatically enters the full OpenSpec workflow (explore → propose → apply).

When to use: The archive directory has accumulated many scattered iteration records and needs periodic cleanup.

How to invoke: Use /lina-archive-consolidate.

lina-auto-archive

Automatically scans the openspec/changes/ directory and batch-archives all completed active changes, then summarizes the results.

The skill strictly checks each change's completion status — only changes where the OpenSpec status is complete and all tasks are done are eligible for archiving. Any incomplete items are skipped; nothing is force-passed or bypassed. After archiving, it reports the list of successfully archived changes and the reasons for any skipped changes.

When to use: Batch-cleaning completed changes or periodically tidying the active changes directory.

Prerequisites: Requires OpenSpec CLI (npm install -g @fission-ai/openspec@latest).

How to invoke: Use /lina-auto-archive.

lina-e2e

Defines the project's Playwright E2E test case management standards: file naming (globally unique TC{NNNN} IDs that are never reused), module-based directory layout, test file templates, independence requirements (one test per file, fully self-contained), and page object model usage.

When to use: Writing, maintaining, or reviewing E2E test cases.

Prerequisites: Install Playwright and its test dependencies first:

cd hack/tests
pnpm install
npx playwright install --with-deps chromium

How to invoke: The AI applies these standards automatically when writing E2E tests, or use /lina-e2e.

lina-feedback

A structured workflow for capturing and resolving issues reported after an OpenSpec change lands. Covers problem classification by type (bug / missing / UX / test gap), incremental spec updates, task organization in tasks.md, fix execution loops, regression analysis, and pre-completion verification.

When to use: After an OpenSpec change has been implemented and feedback reports need systematic triaging and fixing.

How to invoke: Use /lina-feedback and describe the feedback to address.

lina-perf-audit

A comprehensive LinaPro backend API performance audit workflow (manual trigger only). Audits for N+1 queries, missing indexes, unbounded list responses, redundant reads, mergeable SQL calls, blocking operations inside loops, and write SQL executed inside read endpoints.

The audit runs in three phases: Phase 0 (environment setup — database reset, service restart, plugin installation, load-test data seeding), Phase 1 (concurrent sub-agent audits per module), Phase 2 (aggregation — summary report and persistent issue cards under perf-issues/).

Heads up

This skill resets the database, restarts services, and spawns multiple sub-agents. It takes tens of minutes to several hours and consumes significant token budget. Only trigger it when you explicitly need a full performance audit.

When to use: When you explicitly need a systematic performance audit of all backend APIs.

Prerequisites: A running LinaPro dev environment started with make dev.

How to invoke: Explicitly say "run lina-perf-audit" or "run a full backend API performance audit" in the conversation.

lina-review

A structured code and spec review workflow, triggered after OpenSpec apply or feedback tasks complete and before archiving. Covers backend code review (GoFrame conventions, API i18n compliance, distributed cache consistency), RESTful API review, SQL review, E2E test review, and report generation. AGENTS.md is the single source of truth for all review standards.

When to use: After an OpenSpec change has been implemented, as the quality gate before archiving.

How to invoke: Use /lina-review — the AI determines the review scope automatically based on the change.

openspec-explore

Thinking-partner mode for exploring ideas, investigating problems, and clarifying requirements. The AI approaches problems with curiosity and without preconceptions — it reads files, probes the codebase, compares options, draws ASCII diagrams, and surfaces risks and unknowns.

Explore mode is for thinking and investigation, not implementation. It can produce OpenSpec artifacts (proposals, designs) as outputs of the thinking process, but it does not implement production features.

When to use: Before starting a complex task, or whenever you need to think through requirements or architectural options.

How to invoke: Use /openspec-explore and describe what you want to explore.

openspec-propose

Generates a complete OpenSpec change with all required artifacts in a single step: proposal.md, design.md, tasks.md, and any specs/ files. The AI creates artifacts in dependency order to keep design docs and task lists internally consistent.

When to use: Kicking off a new feature or change that needs a full OpenSpec proposal.

How to invoke: Use /openspec-propose and describe the change you want to make.

openspec-apply-change

Implements code, documentation, and test changes according to the tasks.md in an OpenSpec change directory. This skill is typically triggered after the proposal and design are complete, responsible for taking the change from spec to runnable implementation.

When to use: An openspec/changes/<change-id>/ directory already exists and you need to start or continue implementing tasks.

Prerequisites: OpenSpec CLI recommended.

How to invoke: Use /openspec-apply-change, or simply tell the AI to continue implementing the current OpenSpec change.

openspec-archive-change

Archives a completed OpenSpec change by merging incremental specs into the baseline spec and moving the change directory to the archive area.

When to use: The change implementation, review, and verification are all done and it needs to be consolidated into the long-term baseline.

Prerequisites: OpenSpec CLI recommended.

How to invoke: Use /openspec-archive-change, or ask the AI to archive the current completed change.

playwright-cli

A comprehensive Playwright CLI command reference for browser automation: navigating to URLs, clicking, typing, form submission, multi-tab workflows, cookie / localStorage / sessionStorage operations, network route interception, and DevTools integration.

When to use: Automating browser interactions, writing Playwright tests, or debugging browser-based workflows.

Prerequisites: Install Playwright and the required browsers:

cd hack/tests
pnpm install
npx playwright install --with-deps chromium

How to invoke: Describe the browser interactions you need, or use /playwright-cli.