Agent setup
Install an agent of your choice, connect Skills and MCP servers, and start deploying to Cloudflare — all from your editor or terminal.
Terminal-based coding agent that understands your codebase, runs commands, edits files, and manages git. Made by Anthropic.
- Full codebase understanding
- Terminal command execution
- Git operations
- Multi-file editing
Lightweight open-source terminal agent that reads and writes files, runs commands, and browses the web in a sandbox. Made by OpenAI.
- File read/write operations
- Command execution
- Web browsing
- Sandboxed environment
AI-first IDE built on VS Code with multi-file Composer edits and background agents. Made by Cursor.
- Multi-file Composer
- Background agents
- Codebase indexing
- Terminal integration
Editor extension and CLI with agent mode, workspace context, and native PR integration. Made by GitHub.
- Agent mode
- Workspace context
- CLI integration
- PR summaries
Open-source terminal agent with a rich TUI that works with 75+ LLMs. Made by Anomaly.
- 75+ model support
- Rich terminal TUI
- Built-in agents (build/plan)
- LSP integration
Agentic IDE with Cascade context and Flows for multi-step tasks. Made by Codeium.
- Cascade context engine
- Flows automation
- Deep codebase search
- Command suggestions
No agents match this filter.
Compare agents
Capabilities, pricing, and context approaches compared.
| Claude Code | Subscription | Locked | Project memory | |||||
| Codex | Hybrid | Locked | Project memory | |||||
| Cursor | Subscription | Multi-provider | Indexed codebase | |||||
| GitHub Copilot | Subscription | Multi-provider | Indexed codebase | |||||
| OpenCode | BYOK | Multi-provider | Project memory | |||||
| Windsurf | Subscription | Multi-provider | Indexed codebase |
Every agent listed supports Skills and MCP.
Understanding agents
Common types, concepts, and tradeoffs.
Workflow
Where the agent runs changes how you interact with it.
Runs in a shell. Best for automation, scripting, and CI pipelines.
Full code editor with AI first-class. Visual diffs, multi-file edits.
Hosted infrastructure. Ideal for async, long-running work.
Plugs into an existing editor. Lightest install, keeps your setup.
Key concepts
The vocabulary you'll run into when comparing agents.
Reusable prompt packages that teach an agent about a specific domain. Think of them as plugins made of instructions plus slash commands.
The Model Context Protocol — a standard that lets agents call external tools and APIs. Connect an MCP server and the agent knows how to use it.
Which foundation models you can use. Locked supports only the vendor's own models. BYOK (Bring Your Own Key) lets you bring your own API key. Multi-provider supports several providers out of the box.
How the agent retains information about your project. Session only remembers the current conversation. Project memory persists across sessions. Indexed codebase builds a searchable index of your whole repository.
Common tradeoffs
Decisions you'll make when picking an agent.
Cloud agents run on hosted infrastructure and read your code over the network. Local agents run on your own machine, with no code leaving it.
Proprietary agents ship under a closed license you don't control. Open-source agents publish their source under an open license, so you can read, modify, or fork the code.
Locked agents only work with the vendor's own proprietary models. BYOK agents let you bring your own API key and switch between providers and models.
Session context resets when you close the conversation. An indexed codebase is built up front and persists, letting the agent retrieve any file in the repo on demand.