Rules & AGENTS.md
How Cursor's rules, AGENTS.md, user rules, and team rules give Agent persistent project context, and which mechanism to use for which job.
Commands & Shortcuts
Reference for Cursor CLI slash commands, IDE keyboard shortcuts for Agent, Inline Edit and Tab, built-in skills invoked with /, and terminal setup.
Managing Context
How Cursor Agent builds context from search, @-mentions, and rules, how summarization and Max Mode work, and how to keep long sessions productive.
Security & Run Modes
Configure Cursor's Run Modes, Auto-review classifier, sandbox, allowlists, .cursorignore, and CLI permissions to control what the agent can do.
Agent Modes
How Cursor's Agent, Plan, Ask, Debug, and Design modes differ, how to switch between them, and how to pick models, Auto with Cursor Router, and Max Mode.
Cursor CLI
Reference for the Cursor CLI (agent) covering install, auth, every flag and subcommand, headless and CI usage, output formats, Shell Mode, ACP, and config.
Skills
Cursor skills are SKILL.md folders that package workflows, scripts, and references the agent loads on demand, following the open Agent Skills standard.
Subagents
Cursor subagents run delegated work in their own context window, with built-in Explore, Bash, and Browser agents and custom agents defined in .cursor/agents/.
Hooks
Cursor hooks are scripts in hooks.json that run at agent lifecycle events to format, audit, block, or modify what the agent does, independent of model judgment.
MCP
Connect Cursor to external tools with MCP servers via mcp.json, stdio and HTTP transports, OAuth, install links, tool approval, and enterprise allowlists.
Plugins
Cursor plugins bundle rules, skills, subagents, commands, MCP servers, and hooks into packages installed from the Cursor Marketplace or a team marketplace.
Cloud Agents & Automations
Run Cursor agents unattended in cloud VMs, trigger them from schedules and events with Automations, and compose skills, subagents, and hooks into workflows.
Parallel Agents & Worktrees
Run several Cursor agents at once with the Agents Window, isolate each one in a Git worktree, compare models with /best-of-n, and hand long tasks to the cloud.
Bugbot & Agent Review
Use Agent Review for in-editor review of local changes and Bugbot for automated PR review, with BUGBOT.md rules, effort levels, Autofix, and Security Agents.
Debugging
Troubleshoot Cursor when the agent stalls, Tab stops, installs fail, or proxies block AI features, and learn where logs and request IDs live for bug reports.
Tips
Concise Cursor tips grouped by theme, covering prompting, context, modes, review, rules and skills, parallel work, and model choice, from the official docs.