criticalAgent ThreatPrivilege Abuse

Langflow IDOR Authorization Bypass via User-Controlled Flow ID

First seen Jul 8, 2026 · Updated Jul 8, 2026

IDORauthorization-bypasslangflowagent-frameworkflow-executionbroken-access-controlCISA-KEVASI02 · Tool MisuseSurface: Tool LayerPropagation: Single Hop

Langflow, a popular visual builder for LLM/agent workflows, has an authorization bypass that lets an authenticated user execute any other user's flow simply by supplying that victim's flow ID. This is a classic insecure direct object reference (IDOR) that fails to verify flow ownership before execution, allowing cross-tenant abuse in multi-user deployments. CISA has added it to the Known Exploited Vulnerabilities catalog, indicating active exploitation and urgency to patch.

Technical Analysis

The vulnerability lies in Langflow's flow execution endpoint, which accepts a flow ID from the request but does not enforce that the authenticated caller actually owns or has been granted access to that flow before running it. An attacker with any valid authenticated session can enumerate or guess victim flow IDs and trigger execution of flows belonging to other users, inheriting whatever credentials, API keys, connected tools, or data sources those flows reference. This crosses agent/tenant boundaries because flows in Langflow often embed secrets (LLM API keys, database credentials, third-party tool tokens) and orchestrate agentic tool calls, so unauthorized execution can lead to secret exfiltration, unauthorized tool invocation, or lateral movement across a multi-tenant agent platform. Because this is listed in CISA KEV with active exploitation confirmed, the risk extends beyond theoretical IDOR to real-world abuse against internet-facing Langflow instances.

Affected Systems

Langflow

Detection Signatures

  • Anomalous API requests to flow execution endpoints referencing flow IDs not previously associated with the requesting user session
  • High-frequency sequential or randomized flow ID probing patterns in access logs
  • Execution of flows immediately after authentication with no prior flow creation/listing activity by that user
  • Unexpected outbound calls from flows to LLM providers or tools not normally associated with the authenticated user's account
  • Logs showing cross-user flow_id parameters in POST/execute requests

Remediation Steps

  1. 1

    Apply vendor patch immediately

    Update Langflow to the fixed version that enforces flow ownership/authorization checks before execution, per CISA KEV due date.

  2. 2

    Enforce server-side authorization checks

    Ensure all flow execution, read, and modification endpoints validate that the requesting user owns or has explicit sharing permissions for the target flow ID, not just that they are authenticated.

  3. 3

    Rotate embedded secrets

    Rotate API keys and credentials stored within Langflow flows, since they may have been exposed to unauthorized execution.

  4. 4

    Audit execution logs

    Review historical flow execution logs for cross-user flow ID access indicating prior exploitation.

  5. 5

    Restrict network exposure

    Limit Langflow instances to internal networks or VPN access and apply WAF rules to detect IDOR probing until patched.

  6. 6

    Implement per-flow access control lists

    Adopt explicit ACL/RBAC model for flow sharing rather than relying on obscurity of flow IDs.

CVE / Advisory IDs

CVE-2026-55255

Industries Most Exposed

TechnologySaaSAI/ML platform providersEnterprises using low-code agent orchestration

Sources

Respond to this threat

Pro subscribers get a full AI-generated incident-response playbook for this threat — detection, containment, eradication, and recovery steps — plus an unlimited AI Threat Advisor for questions about your environment.