/

CVE-2026-61808

Severity:

CRITICAL RISK

(9.8)

Missing Authentication (CWE-306) in LightRAG API Server

CVE-2026-61808

Deferred

Vulnerability type

Category:

Missing Authentication

CWE:

CWE-306

Impact:

Unauthorized data access, integrity loss, and resource consumption

Severity:

CRITICAL RISK

(9.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

LightRAG API servers through version 1.5.4 may expose administrative and data-management endpoints without authentication while listening on all network interfaces. An unauthenticated network attacker could read indexed documents, alter knowledge-graph data, manage documents and pipelines, clear caches, or consume configured LLM resources.

Affected context

The issue occurs in LightRAG deployments where the API server is reachable over a network and authentication remains disabled by default. Risk is highest when the service binds to externally reachable interfaces without compensating controls such as network isolation, an authenticated reverse proxy, or restrictive firewall policies.

Exploit preconditions

  • Attack surface: Network-accessible LightRAG API endpoints
  • Required access: Network connectivity to the exposed API; valid credentials may not be required when authentication is disabled
  • Trigger point: Requests to document, knowledge-graph, pipeline, cache, or related API operations
  • Why exploitation is possible: Sensitive and state-changing operations can be processed without verifying the caller's identity or permissions

Why this happens

The underlying weakness is a missing authentication control on a network-exposed API. When the server accepts requests without establishing the caller's identity, protected operations become available to any reachable client. This maps to CWE-306: Missing Authentication for Critical Function, particularly where unauthenticated requests can access data or perform administrative and resource-intensive actions.

Potential impact

  • An attacker may read indexed document content
  • An attacker could upload, modify, or delete documents
  • Knowledge-graph data may be altered or poisoned
  • Pipelines may be cancelled or caches may be cleared
  • Configured LLM resources may be consumed, causing cost increases or service degradation
  • Exposed data or service functionality could support further attacks against connected systems

General mitigation guidance

Mitigation generally focuses on enforcing identity and authorization checks before every sensitive API operation and reducing network exposure.

Common defensive patterns include:

  • Require strong authentication by default for API access
  • Apply endpoint-level authorization and least-privilege roles
  • Bind administrative services only to necessary interfaces or place them behind a protected gateway
  • Enforce network segmentation, firewall rules, and private-service access
  • Use rate limits and quotas for resource-intensive operations
  • Separate read-only retrieval capabilities from document-management and administrative functions
  • Avoid exposing secrets, documents, or operational endpoints through unauthenticated routes
  • Monitor and protect high-impact actions with audit logging and alerting

How teams detect issues like this

Organizations can identify this class of weakness through:

  • API inventory reviews that identify services listening on all interfaces
  • Static and configuration analysis for disabled authentication or permissive bind settings
  • Code review checks confirming authentication and authorization middleware covers every sensitive route
  • Dynamic testing that verifies unauthenticated requests are rejected consistently
  • Network scans and exposure assessments for publicly reachable API listeners
  • Policy enforcement requiring authentication, private network placement, and rate limiting for AI service endpoints
  • Monitoring for unusual document changes, graph mutations, cache-clearing actions, pipeline cancellations, or unexpected LLM usage

Sources

SourceLink
github.comgithub.com/HKUDS/LightRAG/commit/0bd102401b4b28a02664e5b6af476bf7a4470292
github.comgithub.com/HKUDS/LightRAG/security/advisories/GHSA-mmg5-8x8q-v934

Frequently asked

questions

Yes, when the LightRAG API is reachable by an attacker and authentication is disabled. The attacker generally needs network connectivity, but does not need an account if the exposed service accepts unauthenticated requests.

The key requirement is network access to the API server. Exposure may come from a public interface, an overly broad internal network route, or a proxy that does not enforce authentication.

Require authentication and endpoint-level authorization by default, restrict the service to trusted networks, and ensure sensitive operations cannot be reached through unauthenticated routes.

Include authentication-coverage tests in CI, maintain an inventory of externally reachable services, enforce secure deployment policies, and monitor high-impact API actions and resource consumption.

Find out what's already
exploitable in your codebase.