Building impactful AI agents requires enabling them to execute business-critical actions. For example, let's take a look at the e-commerce space, where a user might want to cancel an order. An AI agent or chatbot that simply explains how to cancel an order is helpful, but one that can actually cancel an order on behalf of the user is far more impactful.

By nature, these actions are risky, especially when performed by an AI agent. That's why we are building Phantasm. Phantasm allows you to build an impactful AI agent safely by providing an approval layer that a human team can use to monitor and manage the workflow of an AI agent.

With Phantasm, your AI agent workflow will look somewhat like this:

Your AI agent needs to call a function based on an LLM response.

An approval request is sent to the approver.

If an approver approves the request, the function will be executed.

Try Phantasm Today!

If you're building an AI agent and you want to make sure that it can be deployed safely in the real world, please feel free to give it a shot! Phantasm works out of the box and doesn't require you to create an account or any API key.

If you’re interested in our journey and goal, please support us by giving us a star and sharing it within your network. Every bit of support counts!

phantasmlabs
/
phantasm

Toolkits to create a human-in-the-loop approval layer for AI agents to execute business critical actions.

Phantasm offers open-source toolkits that allows you to create human-in-the-loop
(HITL) workflows for modern AI agents. Phantasm comes with 3 main components
that work together to create a seamless HITL experience:

Server: Coordinating the HITL workflows between humans and AI agents.

Dashboard: For the human team to monitor and manage the workflows.

Client: A library to integrate the workflows into your AI agents.

Features

✅ Fully open-source and free to use

✅ Works out of the box with any AI framework or model

✅ Load balancer to distribute the requests to multiple approvers (Beta)

✅ Web-based dashboard to manage the approval workflows

✅ Easy-to-use client libraries for popular programming languages

How It Works

Phantasm allows you to have an approval layer on top of your AI agents. This
means, you're free to use any AI framework or model you see fit. By using
Phantasm, you can delay…

View on GitHub

IR-PLAYBOOK-RCEHIGH

Schritte anzeigen

SOC Incident Playbook: Remote Code Execution (RCE) Defense

WAF MitigationSofort

Virtuelles Patchen am Reverse Proxy / WAF für gefährdete URL-Pfade und Payloads aktivieren.

SecRule REQUEST_URI "@rx ..." "id:1001,phase:1,deny,status:403"

Web Shell ScanSofort

Web-Root auf neu angelegte Scripts, anomal modifizierte Dateien und Persistence scannen.

find /var/www -type f -name "*.php" -mmin -120

Outbound FilteringSofort

Ausgehenden Datenverkehr (Reverse Shells zu unbekannten IPs/Ports) blockieren.

ss -tunp | grep -E ":(4444|1337|8080|9001)"

Vendor Patching

Hersteller-Update in Staging validieren und schnellstmöglich in Produktion ausrollen.

apt update && apt --only-upgrade install <pkg>

1-Click Detection Engineering: Multi-Dialect SIEM Rules5 Formate

Regeln anzeigen

Syntax validiert (0 Fehler)

title: Detect Exploitation - Phantasm: A Human Approval Layer For AI Agents
id: 85a5792a-5c97-4721-8a6e-73c2348a6002
status: experimental
description: Automatisch generierte SIEM-Erkennungsregel basierend auf CTI Intelligence
references:
  - https://tsecurity.de/
author: iShareStuff CTI Automated Detection Engine
date: 2026-09-25
logsource:
  category: network_connection
  product: any
detection:
  selection:
      CommandLine|contains:
        - 'exploit'
  condition: selection
falsepositives:
  - Legitime administrative Zugriffe oder Penetrationstests
level: high
tags:
  - attack.initial_access

Syntax validiert (0 Fehler)

rule CTI_Threat_Indicator {
    meta:
        author = "iShareStuff CTI Automated Detection Engine"
        date = "2026-09-25"
        description = "YARA Signature for "
    strings:
        $str = "Phantasm: A Human Approval Lay" ascii wide
    condition:
        any of them
}

Syntax validiert (0 Fehler)

index=security sourcetype IN ("cisco:asa", "pan:traffic", "zeek_conn", "suricata", "WinEventLog:Security")
("Phantasm A Human Approval Layer For AI A")
| stats count earliest(_time) as first_seen latest(_time) as last_seen by src_ip, dest_ip, dest_host, signature
| eval first_seen=strftime(first_seen, "%Y-%m-%d %H:%M:%S"), last_seen=strftime(last_seen, "%Y-%m-%d %H:%M:%S")
| sort - count

Syntax validiert (0 Fehler)

message: "*Phantasm A Human Approval Layer For AI A*"

Syntax validiert (0 Fehler)

CommonSecurityLog
| where Message has "Phantasm A Human Approval Layer For AI A"
| summarize EventCount = count(), FirstSeen = min(TimeGenerated), LastSeen = max(TimeGenerated) by SourceIP, DestinationIP, DestinationPort, Activity
| extend DetectionRule = "iShareStuff-CTI-Compiled"
| sort by EventCount desc

🎯

MITRE ATT&CK Matrix Navigator 14 Taktiken

Reconnaissance

Resource Development

Initial Access

Execution

Persistence

Privilege Escalation

Defense Evasion

Credential Access

Discovery

Lateral Movement

Collection

Command and Control

Exfiltration

Impact

💡 Tipp: Klicken Sie auf eine TTP, um Gegenmaßnahmen und SIEM-Erkennungsregeln anzuzeigen.Enterprise Matrix v14.1

🚀 Auf tsecurity.de entdecken

◀ VORHERIGER BEITRAG 1 Min

Hacktoberfest 2024: My Open-Source Journey

Sichere Programmierung

NÄCHSTER BEITRAG 1 Min

CVE-2007-5027 | Level One WBR3404TX r1.94p0vtig Web Management DU cross site scripting (EDB-30597 / XFDB-36696)

Sicherheitslücken (CVE)Weiterlesen ▶

Blättern: Pfeiltasten [ ← ] [ → ] · Tasten j / k · mobil: Wischen

Fokus-Vektor:

Kognitive Analyse für identifizierte Bedrohung: Erhöhte Bedrohungslage im Bereich Phantasm: A Human Approval Layer For AI .... Basierend auf 368k Vektor-Korrelationen werden sofortige Isolationsmaßnahmen für betroffene Endpunkte empfohlen.

🛡️ Angriffsfläche & Exposure

Netzwerk/Remote-Zugriff ohne Vorauthentifizierung möglich.

⚡ Empfohlene Sofortmaßnahmen

1. Perimeter-Inspektion: Relevante Portfreigaben und exponierte Endpunkte unverzüglich scannen.
2. Patch-Applikation: Hersteller-Hotfix einspielen oder betroffene Daemons in isolierte DMZ-Segmente überführen.
3. Telemetrie & EDR-Alerts: Prozessaufrufe und Child-Processes auf anomale Shell-Spawns überwachen.

🔗 Semantisch verwandte Zero-Days MariaDB 11.7 VEC

Schnell-Prompts für Analysten

⚠️ DEFANGED SANDBOX VIEW (tsecurity.de): Alle Netzwerkindikatoren und URLs wurden neutralisiert (2 Indikatoren entschärft). Nur zur defensiven Analyse!

Format: python

#!/usr/bin/env python3
# Exploit Title: PoC for Target
import socket, sys
target = '192[.]168[.]1[.]10'
url = 'hxxp://target[.]local/vuln'
s = socket.socket()
s.connect((target, 80))
s.send(b'GET / HTTP/1.1\r\nHost: target.local\r\n\r\n')

Auch interessante Nachrichten Phantasm: A Human Approval Layer For AI Agents

Thematisch verwandte Begriffe: Phantasm, Human, Approval, Layer · 6 Treffer

Sichere Programmierung DEV Community

I found an agent skill that harvests browser credentials. It had 60,000 stars.

vor 4 Tagen 📆 20.09.2026 um 04:52 Uhr

Weiterlesen ↗

Sicherheitslücken (CVE) KitPloit - PenTest Tools!

xalgorix v4.6.78

vor 4 Tagen 📆 19.09.2026 um 21:27 Uhr

Weiterlesen ↗

Sichere Programmierung DEV Community

Orca: The Agent Development Environment for Running AI Coding Agents in Parallel

vor 4 Tagen 📆 20.09.2026 um 08:00 Uhr

Weiterlesen ↗

Sichere Programmierung DEV Community

Hot-Swap a 3D Avatar Without Publishing the Broken Frame

vor 4 Tagen 📆 20.09.2026 um 04:12 Uhr

Weiterlesen ↗

KI & AI Videos

Julian Goldie SEO: I Built a Second Brain for All My AI Agents

Video von Julian Goldie SEO auf YouTube: Get the Agent OS & Obsidian Masterclass 👉 https://www.skool.com/ai-profit-lab-7462/about Want to make money and save…

vor 4 Tagen 📆 20.09.2026 um 00:00 Uhr

Weiterlesen ↗

Sichere Programmierung DEV Community