Subscribe

Hub/Tools/Claude Code

Claude Code

Anthropic · Everything the hub tracks about Claude Code: MCP Clients, AI Plugins, IDE Integrations, AI Security.

MCPPluginsIDEsSecurity39 advisories

MCP Clients

Claude Code

Yes: Tools, Resources, Prompts, Elicitation, Roots, OAuth sign-in, Local (stdio), Remote (HTTP/SSE), Latest protocol

1 of 10 unknown

Full mcp detail →

AI Plugins

Claude Code plugins

Yes: Public directory, MCP-based, Bundles skills, Hooks, Private / custom, Git marketplaces, OAuth sign-in, Org admin controls

1 of 10 unknown

Full plugins detail →

IDE Integrations

Claude Code

Yes: Agent mode, MCP, BYO model / key, VS Code, JetBrains, Zed

5 of 12 unknown

Full ides detail →

AI Security

Claude Code

Yes: Tool-call approval prompts, Per-tool allow / deny, Sandboxing, OAuth for remote MCP, Admin allowlists, Data retention / training opt-out, Security disclosure page

Partial: Auto-install / auto-update control, Audit logs

0 of 10 unknown

Full security detail →

Security advisories (39)

29 high · 4 medium · 3 low · 3 not rated
  1. Not Ratedresolved
  2. High 7.7patched

    TOCTOU Race in Edit Tool Allows File Writes Outside the Workspace

    Claude Code · CVE-2026-88968 · GHSA-5xjx-9qx6-jcrv

    Claude Code
  3. Not Ratedopen
  4. High 7.5patched
  5. High 7.7patched
  6. High 7.7patched
  7. Low 2.0patched
  8. High 8.5patched
  9. Not Ratedresearch
  10. High 7.7patched
  11. Medium 4.4patched
  12. Medium 6.0patched
  13. High 8.5patched
  14. High 7.4patched
  15. High 7.7patched
  16. High 7.7patched
  17. Medium 5.4patched
  18. High 7.7patched

    Workspace Trust Dialog Bypass via Repo-Controlled Settings File

    Claude Code · CVE-2026-33068 · GHSA-mmgp-wc2j-qcv7

    Claude Code
  19. High 7.7patched
  20. High 7.7patched
  21. High 7.7patched

    Command Injection via Directory Change Bypasses Write Protection

    Claude Code · CVE-2026-25722 · GHSA-66q4-vfjg-2qhh

    Claude Code
  22. Low 2.3patched

    Permission Deny Bypass Through Symbolic Links

    Claude Code · CVE-2026-25724 · GHSA-4q92-rfm6-2cqx

    Claude Code
  23. High 7.1patched
  24. High 7.7patched

    Command Injection in find Command Bypasses User Approval Prompt

    Claude Code · CVE-2026-24887 · GHSA-qgqw-h4xq-7w8w

    Claude Code
  25. High 7.7patched
  26. Medium 5.3patched
  27. High 8.7patched

    Command Validation Bypass Allows Arbitrary Code Execution

    Claude Code · CVE-2025-66032 · GHSA-xq4m-mc3c-vvg3

    Claude Code
  28. High 8.7patched

    Sed Command Validation Bypass Allows Arbitrary File Writes

    Claude Code · CVE-2025-64755 · GHSA-7mv8-j34q-vp7q

    Claude Code
  29. High 7.7patched

    Command execution prior to Claude Code startup trust dialog

    Claude Code · CVE-2025-65099 · GHSA-5hhx-v7f6-x7gv

    Claude Code
  30. Low 2.3patched

    Permission deny bypass through symlink

    Claude Code · CVE-2025-59829 · GHSA-66m2-gx93-v996

    Claude Code
  31. High 8.7patched

    Command execution prior to Claude Code startup trust dialog

    Claude Code · CVE-2025-59536 · GHSA-4fgq-fpq9-mr3g

    Claude Code
  32. High 7.7patched
  33. High 8.7patched
  34. High 8.7patched
  35. High 8.7patched
  36. High 7.1patched
  37. High 8.7patched
  38. High 7.7patched
  39. High 8.8patched

Recent updates

  1. Security

    Cursor VS Code Extension Marketplace: Malicious 'Anthropic.claude-code' 2.1.312 / 2.1.313 in Cursor's extension marketplace

    Not Rated. Status: resolved. Cursor says two impersonating versions (Windows / macOS) attempted an SSH backdoor. Cursor removed them, shut down and rebuilt the marketplace, and rotated keys.

    Source

  2. Security

    Claude Code: TOCTOU Race in Edit Tool Allows File Writes Outside the Workspace

    High (CVSS 7.7). Status: patched. CVE-2026-88968 · GHSA-5xjx-9qx6-jcrv. Claude Code advisory.

    Source

  3. Security

    Postman Desktop 12.33.4–12.35.1: Postman Desktop auto-installs Codex / Claude Code plugins and pre-trusts its own startup hook

    Not Rated. Status: open. Users report Postman's auto-existing-upgrade step installing Postman plugins into Claude Code and Codex, writing a trusted hook hash into Codex config, adding a SessionStart hook, the remote MCP server mcp.postman.com and a CLI in ~/.local/bin. Postman staff replied on Nov 3 that opt-in will be made clearer and hook approval is under review. No CVE.

    Source

  4. Security

    Claude Code: Uncontrolled Search Path Enables Arbitrary Code Execution via Malicious Git Executable

    High (CVSS 7.5). Status: patched. CVE-2026-88066 · GHSA-4p2j-qj3c-6x84. Claude Code advisory.

    Source

  5. Security

    Claude Code: Plugin Commit SHA Pin Bypass Allows Installation of Attacker-Controlled Code

    High (CVSS 7.7). Status: patched. CVE-2026-86063 · GHSA-pq7j-f95f-qfcv. Claude Code advisory.

    Source

  6. MCP

    Claude Code defaults to MCP 2026-07-28 everywhere

    Confirmed on 2026-07-28 by default for HTTP, stdio (v2.1.292+) and claude.ai connectors (v2.1.295+). Opt out with MCP_PROTOCOL_NEGOTIATION=legacy.

    Source

  7. Security

    Claude Code: Arbitrary File Write via Write-Time Symlink Following (TOCTOU) in Claude Code

    High (CVSS 7.7). Status: patched. CVE-2026-103435 · GHSA-5j29-h97v-84ch. Claude Code advisory.

    Source

  8. Security

    Claude Code: Managed Settings Not Applied When a Stored API Key Was Preferred Over Enterprise Sign-In

    Low (CVSS 2.0). Status: patched. CVE-2026-103012 · GHSA-gfvf-j8jh-jxxw. Claude Code advisory.

    Source

  9. Security
  10. Security

    MCP Registry (V.A.P.E entry): Shai-Hulud worm payload delivered through an official MCP Registry entry

    Not Rated. Status: research. OX Security reports a registry-listed MCP server shipping a payload that wrote malicious .vscode/settings.json and .claude/settings.json files.

    Source

Other tools

Claude appsChatGPTCodexCursorGitHub CopilotWindsurf / DevinGeminiJetBrains AIKiroAmazon Q DeveloperZedClineRoo CodeContinueGooseReplitPerplexityMicrosoft 365 CopilotTabnineMCP protocol, SDKs and registry

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.

You'll finish signing up on beehiiv.