Security audit

Mailroom

Security checks for vulnerabilities and agentic risk

Overview

Mailroom is a coherent email-routing and approval plugin, but it handles sensitive mailbox, agent-memory, and third-party model data that operators should configure deliberately.

Install only if you are comfortable letting this plugin read and persist Outlook message content and metadata, send selected email context to Gemini, send agent workspace/context/memory evidence to Anthropic during profile generation, and use Telegram as the human approval channel. Configure least-privilege Outlook/Maton access, keep API keys and the SQLite ledger protected, start in shadow mode, and verify the two-step approval flow before production use.

SkillSpector

SkillSpector was not run because this plugin release contains no bundled skills.

Static analysis

Detected: suspicious.dangerous_exec, suspicious.env_credential_access

Shell command execution detected (child_process).

Critical

Code: suspicious.dangerous_exec
Location: dist/cli.js:61
Evidence: execFileSync(cfg.pythonExecutable || "python3", [

Environment variable access combined with network send.

Critical

Code: suspicious.env_credential_access
Location: dist/outlook.js:32
Evidence: const k = process.env.MATON_API_KEY;