# [CRIT] Fake AI Agent Skill Passed Security Scans and Reportedly Reached 26,000 Agents

**Source:** The Hacker News
**Published:** 2026-06-23
**Article:** https://thehackernews.com/2026/06/fake-ai-agent-skill-passed-security.html

## Threat Profile

Fake AI Agent Skill Passed Security Scans and Reportedly Reached 26,000 Agents 
 Swati Khandelwal  Jun 23, 2026 Supply Chain Security / Enterprise Security 
Security firm AIR built a fake AI agent skill, pushed it through a popular skill marketplace and an Instagram ad, and says it reached roughly 26,000 agents, including some on corporate accounts.
Every skill security scanner the firm tested it against marked it safe. The payload was harmless by design: it collected the user's email address …

## Indicators of Compromise (high-fidelity only)

- **CVE:** `CVE-2026-11645`

## MITRE ATT&CK Techniques

- **T1539** — Steal Web Session Cookie
- **T1555.003** — Credentials from Web Browsers
- **T1190** — Exploit Public-Facing Application
- **T1195.002** — Compromise Software Supply Chain
- **T1204.002** — User Execution: Malicious File

## Kill chain phases observed

_(none detected from narrative keywords)_

## Recommended hunts

### Infostealer — non-browser process accessing browser cookie/login DBs

`UC_BROWSER_STEALER` · phase: **actions** · confidence: **High**

**Splunk SPL (CIM):**
```spl
| tstats `summariesonly` count min(_time) as firstTime max(_time) as lastTime
    from datamodel=Endpoint.Filesystem
    where (Filesystem.file_path="*\Google\Chrome\User Data\*\Login Data*"
        OR Filesystem.file_path="*\Google\Chrome\User Data\*\Cookies*"
        OR Filesystem.file_path="*\Microsoft\Edge\User Data\*\Login Data*"
        OR Filesystem.file_path="*\Mozilla\Firefox\Profiles\*\logins.json*"
        OR Filesystem.file_path="*\Mozilla\Firefox\Profiles\*\cookies.sqlite*")
      AND NOT Filesystem.process_name IN ("chrome.exe","msedge.exe","firefox.exe","brave.exe","opera.exe")
    by Filesystem.dest, Filesystem.process_name, Filesystem.file_path, Filesystem.user
| `drop_dm_object_name(Filesystem)`
```

**Defender KQL:**
```kql
DeviceFileEvents
| where Timestamp > ago(7d)
| where InitiatingProcessAccountName !endswith "$"
| where FolderPath has_any (@"\Google\Chrome\User Data\", @"\Microsoft\Edge\User Data\", @"\Mozilla\Firefox\Profiles\")
| where FileName in~ ("Login Data","Cookies","logins.json","cookies.sqlite")
| where InitiatingProcessFileName !in~ ("chrome.exe","msedge.exe","firefox.exe","brave.exe","opera.exe")
| project Timestamp, DeviceName, InitiatingProcessAccountName, InitiatingProcessFileName, FolderPath, FileName, ActionType
```

### Trusted vendor binary / installer launching unusual children

`UC_SUPPLY_CHAIN` · phase: **exploit** · confidence: **Medium**

**Splunk SPL (CIM):**
```spl
| tstats `summariesonly` count min(_time) as firstTime max(_time) as lastTime
    from datamodel=Endpoint.Processes
    where Processes.parent_process_name IN ("setup.exe","installer.exe","update.exe")
      AND Processes.process_name IN ("powershell.exe","cmd.exe","rundll32.exe","regsvr32.exe","mshta.exe","wscript.exe","cscript.exe","wmic.exe","bitsadmin.exe")
    by Processes.dest, Processes.user, Processes.parent_process_name, Processes.process_name, Processes.process
| `drop_dm_object_name(Processes)`
```

**Defender KQL:**
```kql
DeviceProcessEvents
| where Timestamp > ago(7d)
| where AccountName !endswith "$"
| where InitiatingProcessFileName in~ ("setup.exe","installer.exe","update.exe")
| where FileName in~ ("powershell.exe","cmd.exe","rundll32.exe","regsvr32.exe","mshta.exe","wscript.exe","cscript.exe","wmic.exe","bitsadmin.exe")
| project Timestamp, DeviceName, AccountName, InitiatingProcessFileName, FileName, ProcessCommandLine
```

### Article-specific behavioural hunt — Fake AI Agent Skill Passed Security Scans and Reportedly Reached 26,000 Agents

`UC_1_3` · phase: **exploit** · confidence: **High**

**Splunk SPL (CIM):**
```spl
``` Article-specific bespoke detection — Fake AI Agent Skill Passed Security Scans and Reportedly Reached 26,000 Agents ```
| tstats `summariesonly` count earliest(_time) AS firstTime latest(_time) AS lastTime
    from datamodel=Endpoint.Processes
    where (Processes.process_name IN ("skills.sh"))
    by Processes.dest, Processes.user, Processes.process_name,
       Processes.process, Processes.parent_process_name, Processes.process_path
| `drop_dm_object_name(Processes)`
| `security_content_ctime(firstTime)`
| append [
| tstats `summariesonly` count
    from datamodel=Endpoint.Filesystem
    where Filesystem.action IN ("created","modified")
      AND (Filesystem.file_name IN ("skills.sh"))
    by Filesystem.dest, Filesystem.user, Filesystem.process_name,
       Filesystem.file_path, Filesystem.file_name
| `drop_dm_object_name(Filesystem)`
]
```

**Defender KQL:**
```kql
// Article-specific bespoke detection — Fake AI Agent Skill Passed Security Scans and Reportedly Reached 26,000 Agents
// Hunts the actual binaries / paths / commandline fragments named
// in the article instead of a generic technique-class template.
DeviceProcessEvents
| where Timestamp > ago(30d)
| where (FileName in~ ("skills.sh"))
| project Timestamp, DeviceName, AccountName, FileName,
          FolderPath, ProcessCommandLine,
          InitiatingProcessFileName, InitiatingProcessCommandLine
| order by Timestamp desc

// File-creation events for the named binaries / paths
DeviceFileEvents
| where Timestamp > ago(30d)
| where ActionType in ("FileCreated","FileModified")
| where (FileName in~ ("skills.sh"))
| project Timestamp, DeviceName, AccountName, FolderPath,
          FileName, ActionType, InitiatingProcessFileName,
          InitiatingProcessCommandLine
| order by Timestamp desc
```

### IOC-driven hunts (use shared templates)

These are standard IOC-substitution hunts — the canonical SPL and KQL live once in [`_TEMPLATES.md`](../_TEMPLATES.md), so we don't repeat the same boilerplate on every CVE / hash / network-IOC briefing.

- **Asset exposure — vulnerability matches article CVE(s)** ([template](../_TEMPLATES.md#asset-exposure)) — phase: **recon**, confidence: **High**
  - CVE(s): `CVE-2026-11645`


## Why this matters

Severity classified as **CRIT** based on: CVE present, 4 use case(s) fired, 5 technique(s) inferred. Read the full article for actor attribution, tooling details, and any defanged IOCs in the body that aren't visible in the RSS summary.