cursor-ai-chat
Provides best practices for utilizing Cursor's chat and inline code editing interface effectively.
Install
mkdir -p .claude/skills/cursor-ai-chat && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/1599" && unzip -o skill.zip -d .claude/skills/cursor-ai-chat && rm skill.zipInstalls to .claude/skills/cursor-ai-chat
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Master Cursor AI Chat with @-mentions, inline edit, and conversationKey capabilities
- →Add selected code as context to the chat panel
- →Switch AI models during a conversation
- →Add TypeScript types to a function
- →Generate new files based on existing patterns
- →Debug re-render loops in hooks
- →Discuss architecture for real-time notifications
How it works
The skill uses chat panel (Cmd+L) for conversational assistance and inline edit (Cmd+K) for surgical edits, allowing users to provide context via @-symbols and select AI models.
Inputs & outputs
When to use cursor-ai-chat
- →Refactor code using inline edits
- →Explain code behavior in chat
- →Add code context to AI conversations
- →Switch AI models for specific tasks
About this skill
Cursor AI Chat
Master the Cursor AI Chat panel and inline edit for code assistance, debugging, and exploration.
Core Chat Interfaces
Chat Panel (Cmd+L / Ctrl+L)
The side-panel chat for conversational code assistance:
┌─────────────────────────────────┐
│ Chat [M] │ ← Model selector
│─────────────────────────────────│
│ Context Pills: │
│ [main.ts] [utils/] [@Web] │ ← Active context shown as pills
│─────────────────────────────────│
│ User: Explain the auth flow │
│ in @src/auth/middleware.ts │
│ │
│ AI: The middleware checks... │
│─────────────────────────────────│
│ [Type message...] [Send] │
└─────────────────────────────────┘
Key actions:
- Select code in editor, then
Cmd+Lto add selection as context Cmd+Shift+Ladds selection to existing chat without clearing- Click model name in top-right to switch models mid-conversation
Inline Edit (Cmd+K / Ctrl+K)
Surgical edits within the editor. Select code (or place cursor), press Cmd+K, type instruction:
Selected: function calculateTotal(items) { ... }
Prompt: "Add TypeScript types and handle empty array edge case"
Result: Cursor shows diff inline -- green for additions, red for removals.
Press Cmd+Y to accept, Esc to reject.
Best for single-function edits, type annotations, refactoring a specific block, adding error handling.
@-Symbol Reference
Type @ in chat to access context sources:
| Symbol | Purpose | Example |
|---|---|---|
@Files | Reference specific files | @src/utils/auth.ts |
@Folders | Include directory contents | @src/components/ |
@Code | Reference specific symbols | @handleSubmit function |
@Docs | External documentation | @Docs React Router |
@Git | Git diff context | @Git (uncommitted changes) |
@Codebase | Semantic search full codebase | @Codebase where is auth handled? |
@Web | Live web search | @Web latest Next.js 15 changes |
@Definitions | Symbol definitions | Jump to type/function definitions |
@Cursor Rules | Active project rules | Show which rules are loaded |
@Recent Changes | Recent file edits | Context from recent modifications |
@Lint Errors | Current lint issues | Fix active linting problems |
@Docs: Adding Custom Documentation
Register external docs for @Docs lookup:
Cursor Settings>Features>Docs- Click
Add new doc - Enter URL:
https://react.dev/reference/react(Cursor crawls and indexes it) - Reference in chat:
@Docs Reactthen ask questions
Works with any documentation site. Good for framework APIs, internal wikis, design systems.
Effective Prompting Patterns
Pattern 1: Context-First Prompts
@src/api/orders.ts @src/types/order.ts
The createOrder function doesn't validate the shipping address.
Add Zod validation that checks: street, city, state (2-letter), zip (5 or 9 digit).
Throw a typed ValidationError if invalid.
Pattern 2: Reference-Based Generation
@src/api/users.ts
Create a new file src/api/products.ts following the exact same pattern
as the users API. CRUD endpoints, same error handling, same response format.
Product fields: id, name, price (cents), category, createdAt.
Pattern 3: Debug with Context
@src/hooks/useAuth.ts @Lint Errors
The useAuth hook causes an infinite re-render loop when the token expires.
The useEffect fires repeatedly. What's the root cause and how do I fix the
dependency array?
Pattern 4: Architecture Discussion
@src/components/ @src/api/
I need to add real-time notifications. The app uses REST APIs currently.
Should I add WebSockets, SSE, or polling? Consider the existing architecture
shown in these directories.
Multi-Turn Conversation Management
Start new chats for new topics. Long conversations degrade response quality as context fills up.
When to start fresh:
- Switching to a different feature or bug
- After 10+ turns on the same topic
- When responses start repeating or losing accuracy
When to continue:
- Iterating on the same code change
- Follow-up questions about the same file
- Asking for tests after generating code
Model Selection for Chat
Switch models using the dropdown in chat header:
| Task | Recommended Model |
|---|---|
| Quick questions, explanations | GPT-4o, Claude Sonnet |
| Complex refactoring, architecture | Claude Opus, GPT-5 |
| Bug hunting, reasoning | o1, o3 reasoning models |
| Speed-critical simple tasks | cursor-small, GPT-4o-mini |
| Auto (let Cursor decide) | Auto mode |
Chat vs Inline Edit vs Composer
| Feature | Chat (Cmd+L) | Inline (Cmd+K) | Composer (Cmd+I) |
|---|---|---|---|
| Scope | Exploration, Q&A | Single block edit | Multi-file changes |
| Context | Conversation history | Selected code | Project-wide |
| Output | Text + code snippets | Inline diff | File-level diffs |
| Best for | Understanding, planning | Quick fixes, types | Features, scaffolding |
Enterprise Considerations
- Privacy Mode: Enable in Cursor Settings > General > Privacy Mode for zero data retention
- Model governance: Teams can restrict which models developers access via admin dashboard
- Audit: Chat interactions are not logged server-side when Privacy Mode is on
- Cost control: Use Auto mode or default to faster models; reserve Opus/GPT-5 for complex tasks
Resources
When not to use it
- →When a new topic requires a fresh conversation
- →After 10+ turns on the same topic
- →When responses start repeating or losing accuracy
Limitations
- →Long conversations degrade response quality as context fills up
How it compares
This skill provides structured interfaces for AI interaction within a code editor, unlike generic chat interfaces that lack direct code context integration and inline modification capabilities.
Compared to similar skills
cursor-ai-chat side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| cursor-ai-chat (this skill) | 3 | 26d | Review | Beginner |
| godot | 1,044 | 5mo | Review | Intermediate |
| software-architecture | 333 | 6mo | No flags | Intermediate |
| drizzle | 238 | 2mo | No flags | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by jeremylongshore
View all by jeremylongshore →You might also like
godot
bfollington
This skill should be used when working on Godot Engine projects. It provides specialized knowledge of Godot's file formats (.gd, .tscn, .tres), architecture patterns (component-based, signal-driven, resource-based), common pitfalls, validation tools, code templates, and CLI workflows. The `godot` command is available for running the game, validating scripts, importing resources, and exporting builds. Use this skill for tasks involving Godot game development, debugging scene/resource files, implementing game systems, or creating new Godot components.
software-architecture
davila7
Guide for quality focused software architecture. This skill should be used when users want to write code, design architecture, analyze code, in any case that relates to software development.
drizzle
lobehub
Drizzle ORM schema and database guide. Use when working with database schemas (src/database/schemas/*), defining tables, creating migrations, or database model code. Triggers on Drizzle schema definition, database migrations, or ORM usage questions.
screenshot-to-code
OneWave-AI
Convert UI screenshots into working HTML/CSS/React/Vue code. Detects design patterns, components, and generates responsive layouts. Use this when users provide screenshots of websites, apps, or UI designs and want code implementation.
zustand
lobehub
Zustand state management guide. Use when working with store code (src/store/**), implementing actions, managing state, or creating slices. Triggers on Zustand store development, state management questions, or action implementation.
codex
Lucklyric
Invoke Codex CLI for complex coding tasks requiring high reasoning capabilities. This skill should be invoked when users explicitly mention "Codex", request complex implementation challenges, advanced reasoning, or need high-reasoning model assistance. Automatically triggers on codex-related requests and supports session continuation for iterative development.