Hub/AI Security/Per-tool allow / deny
Which AI tools have per-tool allow / deny?
Rules that pre-approve or block specific tools, commands or MCP tools
7 Yes2 Partial
Verified Yes: Claude Code, Codex, Cursor, Windsurf / Devin Desktop, Gemini CLI, Kiro, Zed.
| Tool | Status | Notes | Source |
|---|---|---|---|
| Claude CodeAnthropic | Yes | allow / ask / deny rules, incl. mcp__server__tool patterns | Source |
| CodexOpenAI | Yes | Per-server enabled_tools / disabled_tools and default_tools_approval_mode | Source |
| CursorAnysphere | Yes | Allowlisted commands and tools run without review | Source |
| Gemini CLIGoogle | Yes | includeTools / excludeTools per server | Source |
| KiroAWS | Yes | autoApprove and disabledTools per server | Source |
| Windsurf / Devin DesktopCognition | Yes | Terminal allow / deny rules; per-tool MCP approval | Source |
| ZedZed Industries | Yes | Per-tool rules incl. mcp:<server>:<tool> | Source |
| ClineCline | Partial | Auto-approve per action category, incl. MCP | Source |
| GitHub Copilot (VS Code)GitHub / Microsoft | Partial | Auto-approve settings and edit globs; per-tool page returned 404 | Source |
| ChatGPT (apps & connectors)OpenAI | Unknown | Not found in official docs checked 2026-11-04 | — |
| Claude apps (claude.ai / Desktop)Anthropic | Unknown | Not found in official docs checked 2026-11-04 | — |
| JetBrains AIJetBrains | Unknown | Not found in official docs checked 2026-11-04 | — |
Other AI Security features
Weekly, free
What changed in AI integrations
One short email when MCP clients, plugin ecosystems, IDE integrations, AI plan prices or security advisories change. Every item is sourced.