Severity by source
AV:N/AC:H/PR:L/UI:N/S:C/C:H/I:H/A:L
Network endpoint but attacker needs a valid API key (PR:L) and an unguessable flow UUID (AC:H); crossing into another user's authorization scope justifies S:C, with full data/execution impact and resource consumption (A:L).
Primary rating from Vendor (https://github.com/langflow-ai/langflow).
CVSS VectorVendor: https://github.com/langflow-ai/langflow
CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:C/C:H/I:H/A:L
Lifecycle Timeline
9DescriptionCVE.org
Summary
Insecure Direct Object Reference (IDOR) vulnerability in /api/v1/responses endpoint allows an authenticated attacker to execute any flow belonging to another user by specifying the victim's flow ID in the request.
Details
The vulnerability exists in the get_flow_by_id_or_endpoint_name helper function in src/backend/base/langflow/helpers/flow.py (lines 399-414).
When a flow is accessed via UUID (flow_id), the function queries the database directly without verifying if the authenticated user owns that flow:
# src/backend/base/langflow/helpers/flow.py:399-414
async def get_flow_by_id_or_endpoint_name(flow_id_or_name: str, user_id: str | UUID | None = None) -> FlowRead:
async with session_scope() as session:
try:
flow_id = UUID(flow_id_or_name)
# When using UUID, query directly WITHOUT checking user_id
flow = await session.get(Flow, flow_id)
# ❌ No user_id check!
except ValueError:
endpoint_name = flow_id_or_name
stmt = select(Flow).where(Flow.endpoint_name == endpoint_name)
# Only when using endpoint_name is user_id checked
if user_id:
stmt = stmt.where(Flow.user_id == uuid_user_id)This function is used by the /api/v1/responses endpoint (defined in src/backend/base/langflow/api/v1/openai_responses.py:589).
PoC (Proof of Concept)
# Attacker (user A) with API_KEY_A tries to execute victim (user B)'s flow
curl -X POST "http://localhost:7860/api/v1/responses" \
-H "x-api-key: sk-ATTACKER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "VICTIM_FLOW_ID",
"input_value": "test",
"stream": false
}'
# Returns 200 and executes the victim's flowImpact
Any authenticated user can:
- Execute any flow in the system by knowing its flow ID
- Access potentially sensitive data processed by victim's flows
- Consume victim's resources
Fixes
Fixed in PR #12832 (fix(security): close IDOR in get_flow_by_id_or_endpoint_name), merged 2026-04-22, released in Langflow 1.9.1.
The helper normalizes user_id once and enforces ownership on both lookup branches (UUID *and* endpoint_name):
flow_id = UUID(flow_id_or_name)
flow = await session.get(Flow, flow_id)
if flow is not None and uuid_user_id is not None and flow.user_id != uuid_user_id:
flow = None
# cross-user lookup falls through to the shared 404Key points:
- Cross-user lookups return 404 (not 403), so flow existence is not disclosed via a 403-vs-404 oracle.
/api/v1/responsesand/api/v2/workflowpassuser_idexplicitly, so fixing the helper closes them directly; the/api/v1/run*routes were additionally moved from a bareDepends(get_flow_by_id_or_endpoint_name)to auth-aware wrapper dependencies (defense in depth).- A malformed
user_idnow fails closed (404 instead of a raw 500). - Webhook routes intentionally keep the unscoped lookup (public by design / explicit ownership check elsewhere).
- Regression tests cover the cross-user UUID case and reproduce the original PoC against
/api/v1/responses.
Acknowledgements
Thanks to the security researchers who responsibly disclosed this vulnerability:
- @yzeirnials
- @johnatzeropath
- @LeftenantZero
- @Zwique
Articles & Coverage 2
AnalysisAI
Cross-user flow execution in Langflow (< 1.9.1) lets any authenticated API-key holder run another user's flow by passing the victim's flow UUID as the model value to the /api/v1/responses endpoint, exposing data the victim's flow processes and consuming their resources. The flaw is an Insecure Direct Object Reference (CWE-639) in the get_flow_by_id_or_endpoint_name helper, which skipped ownership checks on the UUID lookup path. Publicly available exploit code exists (a working curl PoC), and the input lists it as confirmed actively exploited (CISA KEV), though the SSVC exploitation status of 'poc' and a low 0.24% EPSS complicate that claim (see risk_assessment).
Technical ContextAI
Langflow is an open-source, Python-based visual builder for LLM/agent workflows ('flows') exposing an OpenAI-compatible responses API. The root cause is CWE-639 (Authorization Bypass Through User-Controlled Key): in src/backend/base/langflow/helpers/flow.py, get_flow_by_id_or_endpoint_name resolved a flow via session.get(Flow, flow_id) whenever the supplied identifier parsed as a UUID, performing no Flow.user_id == caller check; only the fallback endpoint_name branch filtered by user, and even then only when a truthy user_id was passed. Because /api/v1/responses (openai_responses.py:589) and the /run* routes wired this helper as a bare FastAPI Depends, the user_id came from an unset query parameter, so lookups ran unscoped. Affected package is pkg:pip/langflow. The fix (PR #12832) normalizes user_id once, nulls the flow on any cross-user match so a shared 404 fires (avoiding a 403-vs-404 existence oracle), fails closed to 404 on malformed user_id, and moves the /run* routes onto auth-aware wrapper dependencies.
RemediationAI
Vendor-released patch: upgrade Langflow to 1.9.1 or later (EUVD indicates 1.9.2 as a safe floor, so prefer 1.9.2+ to reconcile the version discrepancy); the fix is PR #12832 / commit 2c9f498d664a3c32698b57d7c5e752625291060e, detailed in advisory https://github.com/langflow-ai/langflow/security/advisories/GHSA-qrpv-q767-xqq2. If you cannot patch immediately, restrict network access to the /api/v1/responses, /api/v1/run*, and /api/v2/workflow endpoints to trusted callers (for example via a reverse proxy or network policy) and treat every issued API key as capable of executing any flow - revoke or tightly scope keys and avoid running untrusted tenants on a shared instance, since the platform provides no per-flow ownership enforcement pre-1.9.1. Because flow IDs are UUIDs, avoid logging or exposing them where other tenants can read them, as knowledge of the UUID is the practical exploitation prerequisite; note these controls reduce but do not eliminate risk from any already-authenticated user who has learned a victim flow ID.
Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t
BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser
pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi
The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica
OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph
pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.
Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301
In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse
Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/
pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne
Integer overflow in the get_data function in zipimport.c in CPython (aka Python) before 2.7.12, 3.x before 3.4.5, and 3.
Same technique Authentication Bypass
View allShare
External POC / Exploit Code
Leaving vuln.today
EUVD-2026-38517
GHSA-qrpv-q767-xqq2