Cline reads workspace rules from .clinerules/ or .cline/rules/, global rules from ~/Documents/Cline/Rules, and also detects .cursorrules, .windsurfrules, AGENTS.md and ~/.agents/AGENTS.md. MCP servers are stored in Cline's cline_mcp_settings.json. Auto Approve toggles control reads, edits and commands. You can use a Claude Pro/Max plan through the Claude Code provider (it calls your installed, logged-in Claude CLI) or a ChatGPT plan through the OpenAI Codex sign-in provider.
Cline at a glance
| Global instructions | ✓ Yes~/Documents/Cline/Rules (also ~/.cline/rules), ~/.agents/AGENTS.md |
|---|---|
| Project instructions | ✓ Yes.clinerules/ or .cline/rules/; auto-detects .cursorrules, .windsurfrules, AGENTS.md |
| Nested / subfolder | ? Not documented Not documented |
| AGENTS.md | ✓ YesAGENTS.md and global ~/.agents/AGENTS.md |
| Skills (SKILL.md) | ✓ Yes.cline/skills/ (recommended), .clinerules/skills/, .claude/skills/; global ~/.cline/skills/ |
| MCP config | ✓ Yes IDE: MCP Servers → Configure → cline_mcp_settings.json (mcpServers). CLI: ~/.cline/data/settings/cline_mcp_settings.json |
| Permissions / allowlist | ◐ Partial Auto Approve per category (read, edit, safe commands, all commands) and YOLO mode; the model flags commands as requiring approval, no fixed allowlist |
| Sandbox | ? Not documented Not documented for the local extension |
| Hooks | ◐ Partial Lifecycle hooks via plugins |
| Claude Pro/Max login | ✓ Yes Claude Code provider: uses your Claude Pro/Max limits via the installed, logged-in Claude CLI |
| ChatGPT Plus/Pro login | ✓ Yes OpenAI Codex provider: sign in with your ChatGPT account (OAuth) |
| API key / own plan | ✓ Yes Anthropic, OpenAI, OpenRouter, Bedrock, Gemini and many more providers |
Exact file paths
| Path | Scope | What it does |
|---|---|---|
.clinerules/ or .cline/rules/ | Workspace | Rule markdown files |
~/Documents/Cline/Rules | User | Global rules |
AGENTS.md / ~/.agents/AGENTS.md | Workspace / user | Cross-tool instructions |
.cursorrules / .windsurfrules | Workspace | Detected automatically |
cline_mcp_settings.json | User | MCP servers (CLI: ~/.cline/data/settings/) |
.cline/skills/<name>/SKILL.md | Workspace | Skills (global: ~/.cline/skills/) |
Using a shared AGENTS.md with Cline
Native: AGENTS.md is detected automatically. See one AGENTS.md for every agent, or generate one with the AGENTS.md generator.
Copy-paste examples
Workspace rule
File: .clinerules/testing.md
# Testing
- Run `pytest -q` after every change to `app/`.
- Never edit files under `migrations/` without asking.
Safe permissions preset
Safe Cline Auto Approve settings
File: Cline → Auto Approve
Read project files ............ ON
Read all files ................ OFF (keeps ~/.ssh, ~/.aws out of reach without asking)
Edit project files ............ ON (review diffs before committing)
Edit all files ................ OFF
Execute safe commands ......... ON
Execute all commands .......... OFF (rm -rf, git push --force, psql DROP ... always ask)
Use the browser ............... OFF
Use MCP servers ............... ask per tool
YOLO mode ..................... OFF
Cline has no fixed command allowlist: the model marks each command as safe or requiring approval, so treat "safe commands" as best-effort. Keep "Execute all commands" and YOLO off and add a rule such as "Never run destructive commands; ask first" in .clinerules.
Gotchas
- Command safety is decided by the model's approval flag, not by a deterministic list.
- The Claude Code provider runs your installed Claude CLI and uses its subscription limits; some features (such as images) are limited in that mode.
- Workspace rules win over global rules when they conflict; both are combined.
FAQ
Where are Cline rules stored?
Can Cline use my Claude Max subscription?
Where is the Cline MCP settings file?
Sources
Checked against these official pages on 2026-10-12. Cells marked "Not documented" mean we could not find the behaviour in these docs; it may still exist.