CVE-2025-63390

5.3 MEDIUM

📋 TL;DR

An authentication bypass vulnerability in AnythingLLM v1.8.5 allows unauthenticated attackers to enumerate and retrieve detailed information about all configured workspaces via the /api/workspaces endpoint. This exposes sensitive configuration data including AI model settings, system prompts, and operational parameters. Organizations running vulnerable versions of AnythingLLM with internet-facing deployments are affected.

💻 Affected Systems

Products:
  • AnythingLLM
Versions: v1.8.5
Operating Systems: All
Default Config Vulnerable: ⚠️ Yes
Notes: The vulnerability exists in the default configuration of the affected version.

📦 What is this software?

⚠️ Risk & Real-World Impact

🔴

Worst Case

Attackers could map the entire workspace structure, gather AI model configurations and system prompts, potentially enabling further attacks or intellectual property theft.

🟠

Likely Case

Unauthenticated information disclosure revealing workspace configurations, AI settings, and operational parameters that could aid in targeted attacks.

🟢

If Mitigated

Limited exposure if proper network segmentation and authentication controls are implemented.

🌐 Internet-Facing: HIGH - Unauthenticated remote attackers can directly exploit this vulnerability without any credentials.
🏢 Internal Only: MEDIUM - Internal attackers or compromised systems could still exploit this, but requires network access.

🎯 Exploit Status

Public PoC: ⚠️ Yes
Weaponized: LIKELY
Unauthenticated Exploit: ⚠️ Yes
Complexity: LOW

Simple HTTP GET request to /api/workspaces endpoint without authentication demonstrates the vulnerability.

🛠️ Fix & Mitigation

✅ Official Fix

Patch Version: Check latest version from Mintplex-Labs/anything-llm repository

Vendor Advisory: https://github.com/Mintplex-Labs/anything-llm/issues

Restart Required: Yes

Instructions:

1. Check current version using version command
2. Update to latest version from official repository
3. Restart the AnythingLLM service
4. Verify authentication is now required for /api/workspaces endpoint

🔧 Temporary Workarounds

Network Access Control

linux

Restrict access to AnythingLLM API endpoints using firewall rules or network segmentation

iptables -A INPUT -p tcp --dport [ANYTHINGLLM_PORT] -s [TRUSTED_NETWORK] -j ACCEPT
iptables -A INPUT -p tcp --dport [ANYTHINGLLM_PORT] -j DROP

Reverse Proxy Authentication

all

Implement authentication at reverse proxy level before requests reach AnythingLLM

# Configure nginx with basic auth
location /api/ {
  auth_basic "Restricted";
  auth_basic_user_file /etc/nginx/.htpasswd;
  proxy_pass http://localhost:[ANYTHINGLLM_PORT];
}

🧯 If You Can't Patch

  • Implement strict network access controls to limit exposure to trusted networks only
  • Monitor and alert on unauthorized access attempts to the /api/workspaces endpoint

🔍 How to Verify

Check if Vulnerable:

Send HTTP GET request to http://[TARGET]:[PORT]/api/workspaces without authentication. If it returns workspace data, system is vulnerable.

Check Version:

Check AnythingLLM web interface or configuration for version information

Verify Fix Applied:

Attempt same unauthenticated request to /api/workspaces endpoint. Should receive authentication error or 401/403 response.

📡 Detection & Monitoring

Log Indicators:

  • Unauthenticated requests to /api/workspaces endpoint
  • Multiple GET requests to workspace API from single IP

Network Indicators:

  • Unusual volume of requests to /api/workspaces from external IPs
  • GET requests to /api/workspaces without authentication headers

SIEM Query:

source="web_logs" AND uri="/api/workspaces" AND NOT (auth_token EXISTS OR session_id EXISTS)

🔗 References

📤 Share & Export