mirror of
https://github.com/upstash/context7.git
synced 2026-09-14 19:09:34 +08:00
4663c15552
* feat(cli): adopt .agents/skills as universal install target Install to .agents/skills/ by default as the shared standard directory. Universal agents (Amp, Codex, OpenCode, Gemini CLI, GitHub Copilot) all read from this path natively. Vendor-specific agents (Claude Code, Cursor, Antigravity) get symlinks pointing back to .agents/skills/. - Add "universal" as a first-class IDE type across all path maps - Auto-detect .agents/ and vendor dirs: detected → confirm, nothing detected → checkbox picker with Universal pre-checked - Update list/remove commands to check universal + vendor-specific dirs - Update DEFAULT_CONFIG.defaultIde from "claude" to "universal" - Show labeled install summary (Universal, Claude Code, etc.) with paths * feat(cli): consolidate agent flags into single --universal option Replace --codex, --opencode, and --amp flags with a single --universal flag that points to .agents/skills/. This simplifies the CLI interface since these agents all share the same installation path. * refactor(cli): simplify IDE types and remove UNIVERSAL_AGENTS Set - Simplify IDE type from 7 types to 4 (remove codex, opencode, amp) - Remove UNIVERSAL_AGENTS Set and replace with direct equality checks - Update documentation to add --universal flag examples throughout - Consolidate Supported Clients table to show single Universal row All universal agents (Amp, Codex, Gemini CLI, GitHub Copilot, OpenCode, etc.) are now represented by the single "universal" type since they're functionally identical and use the same .agents/skills/ directory. * docs(skills): update CLI flags to use --universal - Replace --codex, --opencode, --amp flags with single --universal flag - Update all command examples (install, list, remove, suggest, generate) - Consolidate Supported Clients table with Universal at top - Reorder tables to prioritize Universal agents
411 lines
12 KiB
Plaintext
411 lines
12 KiB
Plaintext
---
|
|
title: Skills
|
|
description: Install, search, and manage AI coding assistant skills from the Context7 registry
|
|
---
|
|
|
|
Skills are reusable prompt templates that extend the capabilities of your AI coding assistants. They allow you to share and install common workflows like React best practices, web design guidelines, PDF processing, and more across projects and teams.
|
|
|
|
Context7 maintains a registry of skills at [context7.com/skills](https://context7.com/skills) that can be installed directly into your AI coding assistant with a single command.
|
|
|
|
## What Are Skills?
|
|
|
|
Skills follow the [Agent Skills](https://agentskills.io) open standard — a specification that works across multiple AI coding tools. A skill is a directory containing a `SKILL.md` file with instructions that your AI assistant loads when relevant.
|
|
|
|
**Why use skills?**
|
|
|
|
- **Standardize workflows**: Ensure consistent code reviews, commit messages, or documentation across your team
|
|
- **Share expertise**: Package domain knowledge (e.g., "how we handle authentication") as reusable prompts
|
|
- **Save time**: Install pre-built skills instead of repeatedly explaining the same patterns to your AI assistant
|
|
|
|
**Popular skills:**
|
|
|
|
| Skill | What it does |
|
|
|-------|--------------|
|
|
| `vercel-react-best-practices` | Teaches Claude modern React patterns, Server Components, and performance optimization |
|
|
| `web-design-guidelines` | Guides Claude on UI/UX principles, spacing, typography, and responsive layouts |
|
|
| `pdf` / `docx` / `xlsx` | Enables Claude to read, create, and manipulate Office documents and PDFs |
|
|
| `supabase-postgres-best-practices` | Helps Claude write optimized Postgres queries, RLS policies, and auth flows |
|
|
| `seo-audit` | Lets Claude analyze pages for SEO issues and suggest improvements |
|
|
| `browser-use` | Allows Claude to control browsers for testing and automation |
|
|
|
|
## The Context7 Skills Registry
|
|
|
|
The [Context7 Skills Registry](https://context7.com/skills) is a searchable marketplace of skills indexed from GitHub repositories. Each skill is identified by its repository path (e.g., `/anthropics/skills`) and skill name.
|
|
|
|
When browsing or searching skills, you'll see:
|
|
|
|
| Field | Description |
|
|
|-------|-------------|
|
|
| **Name** | The skill identifier used for installation |
|
|
| **Description** | What the skill does and when to use it |
|
|
| **Install Count** | Number of times the skill has been installed |
|
|
| **Trust Score** | Quality and safety indicator (0-10) |
|
|
|
|
**Quick start:**
|
|
|
|
```bash
|
|
# Search for a skill
|
|
npx ctx7 skills search react
|
|
|
|
# Install a popular skill
|
|
npx ctx7 skills install /vercel-labs/agent-skills vercel-react-best-practices
|
|
```
|
|
|
|
## Trust Scores
|
|
|
|
Every skill in the registry has a **trust score** from 0 to 10 that indicates the reliability and safety of the skill source.
|
|
|
|
| Score | Level | Meaning |
|
|
|-------|-------|---------|
|
|
| 7.0 - 10.0 | High | Verified or well-established source |
|
|
| 3.0 - 6.9 | Medium | Standard community contribution |
|
|
| 0.0 - 2.9 | Low | New or unverified — review before using |
|
|
|
|
Trust scores help you make informed decisions about which skills to install. Higher scores indicate skills from reputable sources with community validation.
|
|
|
|
### Security Features
|
|
|
|
Context7 automatically scans skills for potential security issues:
|
|
|
|
- **Prompt injection detection**: Skills containing potentially malicious instructions are blocked from installation
|
|
- **Blocked skill warnings**: When viewing a repository, you'll see how many skills were blocked due to security concerns
|
|
- **Clear error messages**: If you try to install a blocked skill, you'll receive a specific warning explaining why
|
|
|
|
## Prerequisites
|
|
|
|
Skills require the Context7 CLI. You need Node.js 18 or later.
|
|
|
|
```bash
|
|
# Run directly with npx (no installation required)
|
|
npx ctx7 skills search pdf
|
|
|
|
# Or install globally for faster access
|
|
npm install -g ctx7
|
|
```
|
|
|
|
## CLI Commands
|
|
|
|
### Install Skills
|
|
|
|
Install skills from a repository to your AI coding assistant's skills directory.
|
|
|
|
```bash
|
|
# Interactive selection from a repository
|
|
npx ctx7 skills install /anthropics/skills
|
|
|
|
# Install a specific skill
|
|
npx ctx7 skills install /anthropics/skills pdf
|
|
|
|
# Install multiple skills at once
|
|
npx ctx7 skills install /anthropics/skills pdf commit
|
|
|
|
# Install all skills without prompting
|
|
npx ctx7 skills install /anthropics/skills --all
|
|
```
|
|
|
|
**Target a specific client:**
|
|
|
|
```bash
|
|
npx ctx7 skills install /anthropics/skills pdf --claude
|
|
npx ctx7 skills install /anthropics/skills pdf --cursor
|
|
npx ctx7 skills install /anthropics/skills pdf --universal
|
|
npx ctx7 skills install /anthropics/skills pdf --antigravity
|
|
```
|
|
|
|
**Install globally** (available in all projects):
|
|
|
|
```bash
|
|
npx ctx7 skills install /anthropics/skills pdf --global
|
|
```
|
|
|
|
<Note>
|
|
When installing to multiple clients, the CLI creates the skill files in your primary client's directory and symlinks them to other clients.
|
|
</Note>
|
|
|
|
### Search Skills
|
|
|
|
Find skills across all indexed repositories in the registry.
|
|
|
|
```bash
|
|
npx ctx7 skills search pdf
|
|
npx ctx7 skills search typescript
|
|
npx ctx7 skills search "react testing"
|
|
```
|
|
|
|
Search results display:
|
|
- Skill name and repository
|
|
- Description
|
|
- Install count
|
|
- Trust score
|
|
|
|
You can install directly from search results by selecting a skill interactively.
|
|
|
|
### Suggest Skills
|
|
|
|
Automatically discover relevant skills based on your project's dependencies. The CLI scans your `package.json`, `requirements.txt`, or `pyproject.toml` and suggests matching skills from the registry.
|
|
|
|
```bash
|
|
# Scan current project and suggest skills
|
|
npx ctx7 skills suggest
|
|
|
|
# Suggest and install to a specific client
|
|
npx ctx7 skills suggest --claude
|
|
npx ctx7 skills suggest --cursor
|
|
npx ctx7 skills suggest --universal
|
|
|
|
# Suggest and install globally
|
|
npx ctx7 skills suggest --global
|
|
```
|
|
|
|
**How it works:**
|
|
|
|
1. Scans your project for dependencies (Node.js and Python supported)
|
|
2. Queries Context7 for skills that match your dependencies
|
|
3. Shows results with install count, trust score, and which dependency matched
|
|
4. Lets you select and install skills interactively
|
|
|
|
**Supported dependency files:**
|
|
|
|
| File | Language |
|
|
|------|----------|
|
|
| `package.json` | Node.js (dependencies + devDependencies) |
|
|
| `requirements.txt` | Python |
|
|
| `pyproject.toml` | Python (PEP 621 + Poetry) |
|
|
|
|
<Tip>
|
|
Run `npx ctx7 skills suggest` when starting on a new project to quickly find skills for your tech stack.
|
|
</Tip>
|
|
|
|
### List Installed Skills
|
|
|
|
View skills installed in your project or globally.
|
|
|
|
```bash
|
|
# List all installed skills
|
|
npx ctx7 skills list
|
|
|
|
# List for a specific client
|
|
npx ctx7 skills list --claude
|
|
npx ctx7 skills list --cursor
|
|
npx ctx7 skills list --universal
|
|
|
|
# List globally installed skills
|
|
npx ctx7 skills list --global
|
|
```
|
|
|
|
### Show Skill Information
|
|
|
|
Get details about all available skills in a repository.
|
|
|
|
```bash
|
|
npx ctx7 skills info /anthropics/skills
|
|
```
|
|
|
|
This displays:
|
|
- Available skill names
|
|
- Descriptions
|
|
- Direct URLs
|
|
- Quick install commands
|
|
|
|
### Remove Skills
|
|
|
|
Uninstall a skill from your project or global directory.
|
|
|
|
```bash
|
|
# Remove with interactive prompts
|
|
npx ctx7 skills remove pdf
|
|
|
|
# Remove from a specific client
|
|
npx ctx7 skills remove pdf --claude
|
|
npx ctx7 skills remove pdf --universal
|
|
|
|
# Remove from global directory
|
|
npx ctx7 skills remove pdf --global
|
|
```
|
|
|
|
## Generate Custom Skills
|
|
|
|
Create custom skills tailored to your specific needs using AI. This feature requires authentication.
|
|
|
|
### Authentication
|
|
|
|
```bash
|
|
# Log in (opens browser for OAuth)
|
|
npx ctx7 login
|
|
|
|
# Check your login status
|
|
npx ctx7 whoami
|
|
|
|
# Log out
|
|
npx ctx7 logout
|
|
```
|
|
|
|
### Generate a Skill
|
|
|
|
```bash
|
|
# Start interactive generation
|
|
npx ctx7 skills generate
|
|
|
|
# Generate for a specific client
|
|
npx ctx7 skills generate --cursor
|
|
npx ctx7 skills generate --claude
|
|
npx ctx7 skills generate --universal
|
|
|
|
# Generate as a global skill
|
|
npx ctx7 skills generate --global
|
|
```
|
|
|
|
### Generation Workflow
|
|
|
|
1. **Describe your expertise**: Enter what you want the skill to do (e.g., "OAuth authentication with NextAuth.js best practices")
|
|
2. **Select documentation**: Search and choose relevant libraries to inform the skill
|
|
3. **Answer questions**: Respond to 3 clarifying questions to focus the skill
|
|
4. **Review**: See the generated skill and optionally request changes
|
|
5. **Install**: Save the skill to your selected client(s)
|
|
|
|
<Tip>
|
|
Describe best practices and constraints, not step-by-step tutorials. For example, "TypeScript strict mode patterns" is better than "how to write TypeScript."
|
|
</Tip>
|
|
|
|
## Supported Clients
|
|
|
|
The CLI automatically detects installed AI coding assistants and offers to install skills for them.
|
|
|
|
| Client | Project Directory | Global Directory |
|
|
|--------|-------------------|------------------|
|
|
| Universal (Amp, Codex, Gemini CLI, GitHub Copilot, OpenCode + more) | `.agents/skills/` | `~/.config/agents/skills/` |
|
|
| Claude Code | `.claude/skills/` | `~/.claude/skills/` |
|
|
| Cursor | `.cursor/skills/` | `~/.cursor/skills/` |
|
|
| Antigravity | `.agent/skills/` | `~/.agent/skills/` |
|
|
|
|
**Project vs Global:**
|
|
- **Project skills** (default): Installed in your current project directory, available only in that project
|
|
- **Global skills** (`--global`): Installed in your home directory, available across all projects
|
|
|
|
## Skill File Structure
|
|
|
|
Each skill is a directory containing at minimum a `SKILL.md` file:
|
|
|
|
```
|
|
my-skill/
|
|
├── SKILL.md # Main instructions (required)
|
|
├── templates/ # Optional templates
|
|
├── examples/ # Optional example outputs
|
|
└── scripts/ # Optional executable scripts
|
|
```
|
|
|
|
### SKILL.md Format
|
|
|
|
Skills use YAML frontmatter followed by markdown instructions:
|
|
|
|
```yaml
|
|
---
|
|
name: my-skill
|
|
description: What this skill does and when to use it
|
|
---
|
|
|
|
Instructions for the AI assistant go here.
|
|
|
|
Use markdown formatting, code examples, and clear steps.
|
|
```
|
|
|
|
| Field | Required | Description |
|
|
|-------|----------|-------------|
|
|
| `name` | Yes | Identifier for the skill (lowercase, hyphens allowed) |
|
|
| `description` | Yes | Explains what the skill does — used for discovery |
|
|
|
|
The markdown body contains the actual instructions your AI assistant follows when the skill is invoked.
|
|
|
|
## Command Shortcuts
|
|
|
|
For faster usage, the CLI provides short aliases:
|
|
|
|
| Shortcut | Full Command |
|
|
|----------|--------------|
|
|
| `npx ctx7 si` | `npx ctx7 skills install` |
|
|
| `npx ctx7 ss` | `npx ctx7 skills search` |
|
|
| `npx ctx7 ssg` | `npx ctx7 skills suggest` |
|
|
| `npx ctx7 skills i` | `npx ctx7 skills install` |
|
|
| `npx ctx7 skills s` | `npx ctx7 skills search` |
|
|
| `npx ctx7 skills ls` | `npx ctx7 skills list` |
|
|
| `npx ctx7 skills rm` | `npx ctx7 skills remove` |
|
|
| `npx ctx7 skills gen` | `npx ctx7 skills generate` |
|
|
| `npx ctx7 skills g` | `npx ctx7 skills generate` |
|
|
|
|
## Troubleshooting
|
|
|
|
### Permission Denied
|
|
|
|
If you see permission errors when installing or removing skills:
|
|
|
|
```bash
|
|
# Try with sudo for global installations
|
|
sudo npx ctx7 skills install /anthropics/skills pdf --global
|
|
```
|
|
|
|
Or fix directory permissions:
|
|
|
|
```bash
|
|
sudo chown -R $(whoami) ~/.claude/skills
|
|
```
|
|
|
|
### Skill Blocked Due to Prompt Injection
|
|
|
|
If a skill is blocked, it contains content flagged as potentially malicious. This is a security feature — the skill cannot be installed.
|
|
|
|
```
|
|
Error: This skill contains potentially malicious content and cannot be installed.
|
|
```
|
|
|
|
Consider using an alternative skill or contacting the skill author.
|
|
|
|
### Client Not Detected
|
|
|
|
If your AI coding assistant isn't detected, ensure its configuration directory exists:
|
|
|
|
```bash
|
|
# For Claude Code
|
|
mkdir -p .claude
|
|
|
|
# For Cursor
|
|
mkdir -p .cursor
|
|
```
|
|
|
|
### Authentication Issues
|
|
|
|
If login fails or tokens expire:
|
|
|
|
```bash
|
|
# Clear stored credentials
|
|
npx ctx7 logout
|
|
|
|
# Log in again
|
|
npx ctx7 login
|
|
```
|
|
|
|
## Disabling Telemetry
|
|
|
|
The CLI collects anonymous usage data to help improve the product. To disable telemetry, set the `CTX7_TELEMETRY_DISABLED` environment variable:
|
|
|
|
```bash
|
|
# For a single command
|
|
CTX7_TELEMETRY_DISABLED=1 npx ctx7 skills search pdf
|
|
|
|
# Or export in your shell profile (~/.bashrc, ~/.zshrc, etc.)
|
|
export CTX7_TELEMETRY_DISABLED=1
|
|
```
|
|
|
|
## Next Steps
|
|
|
|
<CardGroup cols={2}>
|
|
<Card title="Browse Skills" icon="magnifying-glass" href="https://context7.com/skills">
|
|
Explore the skills registry
|
|
</Card>
|
|
<Card title="Claude Code Skills Docs" icon="book" href="https://docs.anthropic.com/en/docs/agents-and-tools/claude-code/skills">
|
|
Learn more about how skills work in Claude Code
|
|
</Card>
|
|
<Card title="Agent Skills Spec" icon="code" href="https://agentskills.io">
|
|
Read the open standard specification
|
|
</Card>
|
|
</CardGroup>
|