Sweep.dev
Sweep is an Artificial Intelligence (AI) powered code assistant designed to enhance and streamline coding processes particularly for software developers and tea...
Last verified:
What is Sweep.dev?
Sweep.dev is Sweep is the fastest AI coding assistant built specifically for JetBrains IDEs, enabling developers to write code 10x faster through intelligent next-edit autocomplete and an AI agent. The tool works across all JetBrains IDEs including IntelliJ IDEA, PyCharm, Android Studio, WebStorm, PhpStorm, and Rider as a single plugin. With 40k+ installs and a 4.9-star rating on the JetBrains Marketplace, it's trusted by thousands of professional developers daily.
Key features include next-edit autocomplete that suggests precise code changes in milliseconds using Sweep's custom Tab model, an AI Agent with three modes (Ask, Agent, and Planning) that automatically searches the codebase and applies code changes, and context-aware suggestions powered by entire project indexing. The chat sidebar automatically uses the current open file as context, and users can add additional files through @-mentions or by enabling search mode. Sweep also offers AI code review, remote MCP servers with OAuth support, autocomplete syntax highlighting, and web search/fetch tools.
Sweep is designed for professional developers who use JetBrains IDEs and want AI assistance without switching to other editors like Cursor. It's particularly valuable for developers who previously alt-tabbed between Cursor and their favorite JetBrains IDE for better autocomplete. The tool handles quick refactors, unit tests, bug fixes, technical problem-solving, and complex pair programming tasks. Enterprise teams benefit from SOC 2 compliance, zero data retention, SSO, and Privacy Mode enforcement.
The privacy-first approach uses Sweep's own LLMs for unmatched price, performance, and security with no code retained by third parties. Developers can checkpoint and rollback to previous states, revert changes individually or in bulk, and create custom rules through SWEEP.md files for code style preferences, naming conventions, and codebase structure documentation.
Sweep.dev pricing
Pricing model: Freemium
Sweep offers a free trial with 1,000 Autocompletes and $5 API credits included. Paid plans include: Basic at $10/mo for unlimited autocomplete (best for solo developers); Pro at $20/mo with more API credits and priority support (ideal for power users); Ultra at $60/mo with maximum API credit allocation (for heavy agent users). All paid plans receive unlimited autocomplete suggestions and the plan cost as Sweep API Credits. For Agent usage, charges are the underlying LLM cost plus 5% markup. Unused credits roll over for 30 days. Team plan includes $40 API credits per seat with SSO, usage visibility, and Privacy Mode enforcement. Special pricing available for students, educators, and open source maintainers.
Sweep.dev pros
- Next-edit autocomplete suggests precise changes in milliseconds
- Works across all JetBrains IDEs with one plugin
- #1 rated AI plugin on JetBrains Marketplace with 4.9 stars
- 40k+ installs trusted by thousands of professional developers
- AI Agent with Ask, Agent, and Planning modes
- Context-aware suggestions from entire project indexing
- Chat sidebar automatically uses current open file as context
- Checkpoints allow rollback to previous states easily
- Zero data retention with SOC 2 compliance
- Own LLMs ensure better price, performance, and security
- File @-mentions add specific files to chat context
- Planning mode generates reviewable plans before execution
- Diff modal shows and allows rejecting changes before accepting
- SWEEP.md custom rules for code style and preferences
- AI code review for diffs between branches
- Remote MCP servers with full OAuth 2.0/2.1 support
- Autocomplete syntax highlighting across all JetBrains IDEs
- Web search and fetch tools to browse from Sweep
- Terminal integration for running git and lint commands
- Auto Accept option for comfortable users
Sweep.dev cons
- Limited autocomplete uses on free trial (1,000 only)
- 5% markup on pay-as-you-go API credits beyond plan
- Free tier has limited Chat and Agent usage
- Bash tool functionality requires IntelliJ 2025.2+ for Reworked Terminal
- PowerShell used instead of bash on Windows/Linux terminals
- Agent won't always get what you want, requires iteration
- Previous messages in thread can hurt model intelligence
- Need one chat per task for best results
- Self-hosting only available on Enterprise plan
- Higher cost than GitHub Copilot ($20 vs $20 per user)
Frequently asked questions about Sweep.dev
What is Sweep and what does it do?
Sweep is the fastest AI coding assistant built specifically for JetBrains IDEs. It lets you write code 10x faster through intelligent next-edit autocomplete that suggests precise code changes in milliseconds, and an AI Agent that automatically searches your codebase and applies code changes. You can work with your codebase using natural language in the chat sidebar, which automatically uses the current open file as context.
Which JetBrains IDEs does Sweep support?
Sweep works with all JetBrains IDEs including IntelliJ IDEA, PyCharm, Android Studio, WebStorm, PhpStorm, Rider, and more. It's one plugin that works across every JetBrains IDE, with a 4.9-star rating and 40k+ installs on the JetBrains Marketplace.
How do I open Sweep Chat and Agent?
You can open Sweep using ⌘ J (Mac) or Ctrl J (Windows/Linux), or by clicking the Sweep icon in the sidebar. For new chats, use ⌘ N (Mac) or Ctrl N (Windows/Linux). To toggle between Normal and Search mode, use ⌘ . (Mac) or Ctrl . (Windows/Linux).
What are the three Agent modes in Sweep?
Sweep supports three modes: Ask searches the codebase for context and doesn't edit your code; Agent automatically searches the codebase and applies code changes; Planning has the LLM create a plan for you to review and iterate on, which can then be executed in Agent mode. Press shift + tab in agent mode to enable planning mode.
How does Sweep's next-edit autocomplete work?
Sweep's custom Tab model suggests precise code changes in milliseconds using predictive editing. It uses next-edit autocomplete that understands your current workspace and file context. You just hit Tab to accept suggestions. This feature is available across all major editors including JetBrains, VS Code, and Zed.
Is Sweep privacy-safe for enterprise use?
Yes, Sweep is privacy-first and enterprise-ready with SOC 2 compliance and zero data retention. No code is retained by third parties. Sweep uses their own LLMs for unmatched price, performance, and security. The team plan includes Privacy Mode enforcement team-wide, SSO, and usage visibility.
How do I add files to Sweep Chat context?
Use @-mentions to quickly reference and add files to your conversation. For example, typing @auth.py will add auth.py to the chat context. You can also highlight specific code and press ⌘+shift+J (Mac) or Ctrl+shift+J (Windows/Linux) to add it to the context, or enable search mode using ⌘ . (Mac) or Ctrl . (Windows/Linux).
What are checkpoints and how do I use them?
Checkpoints let you rollback to previous states, giving freedom to experiment without losing progress. In each Agent chat, Sweep applies changes as a diff modal that you can view and reject. Click 'Revert Changes' to revert changes in bulk or one at a time. All changes since the chat was sent are tracked.
What is SWEEP.md and how do I use it?
SWEEP.md is a file you create in your project root to provide custom rules and context for Sweep. It stores frequently used commands (build, test, lint), defines code style preferences (naming conventions, libraries, formatting), documents codebase structure, and provides domain-specific context. Sweep automatically reads and follows SWEEP.md guidelines when making changes. If no SWEEP.md exists, CLAUDE.md is used instead.
What's included in the free trial vs paid plans?
The free trial includes 1,000 Autocompletes and $5 API credits included. Basic ($10/mo) offers unlimited autocomplete for solo developers. Pro ($20/mo) includes more API credits and priority support for power users. Ultra ($60/mo) provides maximum API credit allocation for heavy agent users. All paid plans receive unlimited autocomplete and the plan cost as API credits. Agent usage charges the LLM cost plus 5% markup.