Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
9 detections
Filters
Last updated
All Time
Detection languages
3
3
3
Contributors
3
3
2
1
Categories
6
4
1
1
1
Platforms
8
6
6
Products / Services
2
1
1
1
1
MITRE Techniques
6
4
3
2
2
CVEs
71
71
64
58
49
Detects anomalous process spawning behavior initiated by AI IDE extensions (e.g., Amazon Q, Claude Code, Cursor) when loading Model Context Protocol (MCP) configuration files. The rule flags instances where these IDE components trigger suspicious child processes or command-line executions, which could be indicative of remote code execution (RCE) via malicious mcp.json workspace configurations.
Detects anomalous child process execution (e.g., bash, sh, npx, node) spawned directly by AI coding assistants or IDE extensions (Claude Code, Cursor, Amazon Q). This behavior, specifically involving auto-executed configurations like mcp.json, is indicative of exploitation attempts targeting AI coding environments where remote code execution (RCE) can be achieved via untrusted configurations or injected prompts.
Detects the creation or modification of a Claude Code settings file (`.claude/settings.json`). This file can be configured to automatically approve Model Context Protocol (MCP) servers by setting `enableAllProjectMcpServers` to `true` or by using `enabledMcpjsonServers`. This bypasses security prompts designed to warn users about potential command execution, which can lead to Remote Code Execution (RCE) as described in CVE-2025-59536.
Detects network connections from the Claude Code CLI ('claude' or 'claude.exe') to a destination other than the official 'api.anthropic.com' endpoint. This indicates that the 'ANTHROPIC_BASE_URL' environment variable has likely been overridden by a malicious '.claude/settings.json' configuration file in an attempt to intercept and exfiltrate API keys, as described in CVE-2026-21852.
This detection identifies the execution of suspicious shell commands (such as download cradles, reverse shells, or unexpected GUI applications like Calculator) spawned as child processes of the Claude Code CLI ('claude'). This activity is indicative of potential Remote Code Execution (RCE) through the exploitation of malicious hooks in `.claude/settings.json` or malicious MCP server commands, as described in CVE-2025-59536.
Detects suspicious Claude Code configuration keys linked to automatic hook execution or automatic MCP server enabling, which were used to bypass consent and execute commands.
Detects creation or modification of repository controlled Claude Code project configuration files (.claude/settings.json or .mcp.json) which were abused for code execution and token exfiltration.
Detects Claude Code spawning suspicious child processes, consistent with abuse of project configuration hooks or MCP initialisation commands.
Detects the Claude CLI or Claude Code spawning suspicious child processes such as shells or network utilities. This behavior indicates exploitation of malicious hooks or MCP server configurations executing arbitrary commands, as seen in CVE-2025-59536 and CVE-2026-21852.
