The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides installation commands that download and immediately execute remote scripts from an untrusted source.
File: SKILL.md
Evidence: curl -sSL https://raw.githubusercontent.com/brennhill/Kaboom-Browser-AI-Devtools-MCP/STABLE/scripts/install.sh | bash (macOS/Linux)
Evidence: irm https://raw.githubusercontent.com/brennhill/Kaboom-Browser-AI-Devtools-MCP/STABLE/scripts/install.ps1 | iex (Windows/PowerShell)
[COMMAND_EXECUTION]: The documentation includes manual setup instructions requiring the execution of local shell commands to modify file permissions and move binaries.
File: SKILL.md
Evidence: chmod +x kaboom and mv kaboom /usr/local/bin/.
[PRIVILEGE_ESCALATION]: The manual installation steps suggest moving files to /usr/local/bin/, a system-protected directory that typically requires administrative or root privileges.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted browser data including console logs, network response bodies, and DOM snapshots, which can contain hidden instructions targeting the agent.
Ingestion points: Captured logs via kaboom_console_get, network content via kaboom_network_get, and page elements via kaboom_dom_snapshot (all in SKILL.md).
Boundary markers: Absent. The skill does not implement delimiters or warnings to prevent the agent from following instructions embedded within the captured data.
Capability inventory: The skill provides extensive automation tools such as kaboom_browser_click, kaboom_browser_type, kaboom_browser_navigate, and kaboom_browser_upload (all in SKILL.md).
Sanitization: Absent. The captured data is returned to the agent without mention of filtering or escaping.

kaboom-browser-ai-devtools