Reference

Tools & scanners

SentinelLocal ships 13 deterministic scanners and three agent front-ends. Findings carry a Severity and human-readable summary, and can be emitted as pretty CLI, JSON, or SARIF v2.1.0.

Deterministic scanners

ToolDescription
scan_file Static analysis of executables/scripts using goblin (PE/ELF/Mach-O), with hashing, suspicious-import detection, URL/IP extraction.
yara_scan Pure-Rust YARA matching via yara-x; ships a starter ruleset.
scan_processes Cross-platform process snapshot (sysinfo) with stalkerware/RAT name matching.
scan_network TCP/UDP socket enumeration (netstat2) with PID attribution and outbound-connection flagging.
audit_persistence Autostart auditor: systemd / cron / launchd / Windows Run keys.
audit_scheduled Time-based job auditor: Windows Scheduled Tasks, systemd timers, cron, launchd.
audit_hosts_file DNS-hijack detector for /etc/hosts and the Windows hosts file.
scan_browser_extensions Chrome / Edge / Brave / Firefox extension auditor — flags broad host & web-request permissions.
recent_files Surfaces files modified in the last N hours under temp / Downloads / cwd.
ioc_lookup Hashes files and matches them against a local IOC list (drop in MalwareBazaar / abuse.ch / your SOC's feed).
audit_logs Reads recent OS log entries via journalctl (Linux), Get-WinEvent (Windows), or log show (macOS); flags known-bad event IDs, error/fault levels, ssh/sudo failures.
audit_shell_history Scans bash / zsh / fish / PowerShell PSReadLine history for attacker patterns: curl-pipe-to-shell, IEX stagers, base64 decode-and-exec, reverse shells, credential exfiltration.
monitor_snapshot Re-snapshots processes + sockets at a fixed interval and reports the delta — catches short-lived droppers and beaconing connections a single snapshot would miss.

Agent front-ends

CommandDescription
agent One-shot ReAct loop: the model plans, then calls the tools above in response to a natural-language goal.
chat Interactive terminal REPL — a persistent planning conversation with live plan / reasoning / tool / finding output.
serve Local HTTP chat UI for testing the agent interactively in a browser.

Listing tools at runtime

sentinel tools lists every registered tool with its description. Because tool descriptors are converted to OpenAI-style function definitions on the fly, anything registered is automatically available to the agent — see Contributing for how to add your own.