VAITP Dataset

← Back to the dataset

CVE-2026-115419

Improper isolation of conversation state allows cross‑tenant data leakage.

  • CVSS 8.6
  • CWE-488
  • Authentication, Authorization, and Session Management
  • Remote

The A2A server component in Amazon Strands Agents SDK for Python before 1.43.0 and Amazon Strands Agents SDK for TypeScript before 1.5.0 does not properly isolate conversation state between callers. An authenticated remote user could read other tenants' conversation history and add persistent instructions that alter model behavior for subsequent callers via HTTP requests to the A2A endpoint using a different context_id. To mitigate this issue, users should upgrade to strands-agents 1.43.0 (Python) or @strands-agents/sdk 1.5.0 (TypeScript).

CVSS base score
8.6
Published
2026-10-22
OWASP
A01 Broken Access Control
Orthogonal defect classification
Interface
Code defect classification
Incorrect Interface
Category
Authentication, Authorization, and Session Management
Subcategory
Privilege Escalation
Accessibility scope
Remote
Impact
Unauthorized Access
Affected component
strands-agents
Fixed by upgrading
Yes

Solution

Upgrade to strands‑agents ≥ 1.43.0 for Python or @strands‑agents/sdk ≥ 1.5.0 for TypeScript.

Vulnerable code sample

import threading
from strands_agents.sdk import A2AClient

# Global store that incorrectly mixes conversations across tenants
conversation_store = {}
store_lock = threading.Lock()

def handle_request(request):
    """
    Process an incoming HTTP request to the A2A endpoint.
    `request` is a dict with keys: 'tenant_id', 'context_id', 'payload'.
    """
    client = A2AClient()  # authenticated client for the caller
    # VULNERABLE: mixes conversation state across tenants using a shared dict
    with store_lock:
        conv = conversation_store.setdefault(request['context_id'], [])
        conv.append(request['payload'])
        conversation_store[request['context_id']] = conv

    # Forward the accumulated conversation to the remote model
    response = client.send_conversation(conv)
    return response

Patched code sample

import threading
from strands_agents.sdk import A2AClient

# Per-tenant store to isolate conversation state
conversation_store = {}
store_lock = threading.Lock()

def handle_request(request):
    """
    Process an incoming HTTP request to the A2A endpoint.
    `request` is a dict with keys: 'tenant_id', 'context_id', 'payload'.
    """
    client = A2AClient()  # authenticated client for the caller
    # FIX: isolates conversation state by including tenant_id in the store key
    with store_lock:
        tenant_key = f"{request['tenant_id']}:{request['context_id']}"
        conv = conversation_store.setdefault(tenant_key, [])
        conv.append(request['payload'])
        conversation_store[tenant_key] = conv

    # Forward the accumulated conversation to the remote model
    response = client.send_conversation(conv)
    return response

Published fix

Public sources link this CVE to the changes below. The code samples above were generated to illustrate the defect; these links lead to the code the project actually changed.

  • fix(a2a): introduce agent factory for isolating agent context from different callers (#2628) (strands-agents/harness-sdk)
    strands-py/src/strands/multiagent/a2a/executor.py, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor.cancel, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor._execute_streaming, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor._handle_agent_result, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor._handle_streaming_event, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor.__init__, strands-py/src/strands/multiagent/a2a/server.py, strands-py/src/strands/multiagent/a2a/server.py · A2AServer.__init__, strands-py/src/strands/multiagent/a2a/__init__.py
  • fix(a2a): isolate conversation state per context in TypeScript (#2696) (strands-agents/harness-sdk)
    strands-py/src/strands/multiagent/a2a/executor.py, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor._execute_streaming, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor._restore_state, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor._run_with_shared_agent, strands-py/src/strands/multiagent/a2a/executor.py · StrandsA2AExecutor.__init__

Cite this entry

@misc{vaitp:cve2026115419,
  title        = {{Improper isolation of conversation state allows cross‑tenant data leakage.}},
  author       = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
  year         = {2026},
  note         = {VAITP Python Vulnerability Dataset, entry CVE-2026-115419},
  howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2026-115419/}}
}

Download this entry as JSON Download the whole dataset

Introducing the "VAITP dataset": a specialized repository of Python vulnerabilities and patches, meticulously compiled for the use of the security research community. As Python's prominence grows, understanding and addressing potential security vulnerabilities become crucial. Crafted by and for the cybersecurity community, this dataset offers a valuable resource for researchers, analysts, and developers to analyze and mitigate the security risks associated with Python. Through the comprehensive exploration of vulnerabilities and corresponding patches, the VAITP dataset fosters a safer and more resilient Python ecosystem, encouraging collaborative advancements in programming security.

The supreme art of war is to subdue the enemy without fighting.

Sun Tzu – “The Art of War”

:: Shaping the future through research and ingenuity ::

Before you use the VAITP AI models

Please read and accept the following disclaimer:

By choosing "I understand and accept", you acknowledge that you understand and agree to these terms.