CVE-2026-126681: Prompt injection flaw in AWS Labs Timestream for InfluxDB MCP Server leaks API tokens

A prompt injection vulnerability in the AWS Labs Timestream for InfluxDB MCP Server allowed remote unauthenticated attackers to steal configured InfluxDB API tokens by injecting malicious URLs into the MCP client context.

Christopher Ort
By Christopher Ort · 3 min read
Share
A developer's laptop screen showing terminal windows with security warnings and database connection logs, photographed from over the shoulder in a dimly lit office
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What happened
  4. What you should do
  5. What it means
  6. Sources
i10X Quick Take

Quick Take

  1. What: CVE-2026-126681, a prompt injection vulnerability in the AWS Labs Timestream for InfluxDB MCP Server (awslabs.timestream-for-influxdb-mcp-server) before version 0.0.19.
  2. Impact: Remote unauthenticated attackers could inject crafted InfluxDB URLs via prompt injection, causing the server to send its configured InfluxDB API token to an attacker-controlled endpoint.
  3. Who is affected: Users running versions 0.0.12 through 0.0.18 of the MCP server.
  4. What to do: Upgrade to version 0.0.19 or later and rotate any InfluxDB API tokens that were configured for the server.
  5. CVSS score: 5.5 (Medium severity).

In plain English

The AWS Labs Timestream for InfluxDB MCP Server is a tool that lets AI assistants manage Amazon Timestream databases and read or write InfluxDB data. Researchers found a security flaw where an attacker could trick the AI assistant into connecting to a fake database server by sneaking a malicious web address into the conversation. When the MCP server tried to connect, it would accidentally send the real database password (API token) to the attacker’s server instead of the real one.

What you should do: If you use this tool, you should update to the latest version and change your database passwords. If you don’t use this specific developer tool for managing InfluxDB databases, you can ignore this issue.

What happened

AWS Labs published a security bulletin on November 16, 2026, disclosing CVE-2026-126681 in the Timestream for InfluxDB MCP Server. The vulnerability stems from improper restriction of intended endpoints in the server’s connection configuration handling.

The MCP server, which enables Model Context Protocol (MCP) clients such as AI assistants to manage Amazon Timestream for InfluxDB resources, did not validate which endpoints it connects to. A remote unauthenticated attacker could exploit this by using prompt injection to introduce a crafted InfluxDB URL into the MCP client context. The server would then connect to the attacker-controlled URL and transmit its configured InfluxDB API token to that endpoint.

This exposure could allow the attacker to access the customer’s InfluxDB data with the permissions granted to the leaked token. The vulnerability is tracked as CWE-923 (Improper Restriction of Communication Channel to Intended Endpoints) and received a CVSS v3.1 score of 5.5 (Medium severity) with the vector string CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:N.

Versions 0.0.12 through 0.0.18 are affected. AWS Labs addressed the issue in version 0.0.19.

What you should do

AWS Labs recommends that users upgrade to awslabs.timestream-for-influxdb-mcp-server version 0.0.19 or later. Users should also ensure any forked or derivative code is patched to incorporate the fixes.

Critically, customers who used an affected version must rotate any InfluxDB API tokens that were configured for the server, as these tokens may have been exposed.

According to the advisory, there is no workaround that fully fixes this issue. Until upgrading, users should use an InfluxDB API token with the least permissions needed and avoid using the server with MCP clients that process untrusted content.

The patched version 0.0.19 can be installed via pip: pip install awslabs.timestream-for-influxdb-mcp-server==0.0.19

What it means

This vulnerability highlights the security risks inherent in AI assistant integrations, particularly around prompt injection attacks. The flaw demonstrates how malicious input to an AI system can be leveraged to manipulate backend services into leaking sensitive credentials.

The attack vector requires user interaction (the AI assistant must process the injected prompt), but no authentication or special privileges are needed. The confidentiality impact is high, as API tokens provide direct access to database operations.

AWS Labs credited security researchers @Yuremin, @FORIMOC, and @invoke1442 for collaborating on this issue through the coordinated vulnerability disclosure process.

Organizations using MCP servers or similar AI-integrated tools should review their security posture around credential handling and endpoint validation, particularly in systems that process potentially untrusted user input through AI assistants.

It is the second AWS Labs MCP server advisory today, after the path traversal flaw in the AWS Pricing MCP Server, and follows the AWS API MCP Server policy bypass. New cases are tracked in our AI Security hub.

Sources

AWS Security Bulletin 2026-185-AWS; GitHub Security Advisory GHSA-2f9r-g4f3-wc2j (awslabs/mcp); NVD entry CVE-2026-126681; PyPI release awslabs.timestream-for-influxdb-mcp-server 0.0.19.

Christopher Ort

Christopher Ort

Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He is the author of “The Style Penalty”, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.

Editor's picks

  1. CVE-2026-126681: Prompt injection flaw in AWS Labs Timestream for InfluxDB MCP Server leaks API tokens
  2. CVE-2026-115563: OAuth credential leak in Model Context Protocol Python SDK
  3. US and Nine Allies Sign San Francisco Declaration to Protect AI Supply Chains
  4. Apple Launches Health AI Study With OpenEvidence for Up to 200,000 iPhone Users
  5. AMD Completes $8.2 Billion World Labs Acquisition, Fei-Fei Li Becomes Chief Scientist