By participating in this program, you agree to:

• Respect the Community Code of Conduct
• Respect the Intigriti Terms and Conditions
• Respect the scope of the program
• Not discuss or disclose vulnerability information without prior written consent (including PoC's on YouTube and Vimeo)

Validation times

We will validate all submissions within the below timelines, once your submission has been verified by Intigriti.

Introduction

LangChain deeply values the security community's efforts to find vulnerabilities in our products and services. This program covers the products, platforms, and services developers use to build, deploy, and monitor applications powered by large language models.

What we're looking for

We're most interested in findings that reflect a realistic attack scenario with genuine, demonstrable impact on our products or their users. Our key focus areas are:

• Cross-tenant vulnerabilities, where one customer or tenant can access another's data or resources, such as broken tenant isolation, broken access controls, or authorization bypass.
• Server-side vulnerabilities in our applications, APIs, or infrastructure, such as remote code execution, server-side request forgery (SSRF), path traversal, or privilege escalation.
• Agent vulnerabilities, such as escaping the agent execution or sandbox environment, or abusing agent capabilities to reach data, systems, or resources beyond their intended boundaries.

Application

• Pre-Auth Account takeover/OAuth squatting
• Self-XSS that can't be used to exploit other users
• Verbose messages/files/directory listings without disclosing any sensitive information
• CORS misconfiguration on non-sensitive endpoints
• Missing cookie flags
• Missing security headers
• Cross-site Request Forgery with no or low impact
• Presence of autocomplete attribute on web forms
• Reverse tabnabbing
• Open redirects without additional security impact
• Bypassing rate-limits or the non-existence of rate-limits.
• Best practices violations (password complexity, expiration, re-use, etc.)
• Clickjacking without proven impact/unrealistic user interaction
• CSV Injection
• Sessions not being invalidated (logout, enabling 2FA, etc.)
• Tokens leaked to third parties
• Anything related to email spoofing, SPF, DMARC or DKIM
• Content injection without being able to modify the HTML
• Username/email enumeration
• Email bombing
• HTTP Request smuggling without any proven impact
• Homograph attacks
• Banner grabbing/Version disclosure
• Not stripping metadata of files
• Same-site scripting
• Subdomain takeover without taking over the subdomain
• Arbitrary file upload without proof of the existence of the uploaded file
• Blind SSRF without proven business impact (pingbacks aren't sufficient)
• Disclosed/misconfigured Google Maps API keys
• Host header injection without proven business impact

General

• In case that a reported vulnerability was already known to the company from their own tests, it will be flagged as a duplicate
• Theoretical security issues with no realistic exploit scenario(s) or attack surfaces, or issues that would require complex end user interactions to be exploited
• Prompt injection and unsafe model behavior without demonstrated exploitability
• Vulnerabilities in third-party services or software that LangChain does not operate, unless impact can be demonstrated on a LangChain-controlled asset
• Reports generated primarily by AI, automated tools, or scanners without manual validation and a working proof of concept
• Spam, social engineering and physical intrusion
• DoS/DDoS attacks or brute force attacks
• Vulnerabilities that only work on software that no longer receive security updates
• Attacks requiring physical access to a victim's computer/device, man in the middle or compromised user accounts
• Recently discovered zero-day vulnerabilities found in in-scope assets within 14 days after the public release of a patch or mitigation may be reported, but are usually not eligible for a bounty
• Reports that state that software is out of date/vulnerable without a proof-of-concept

This program follows Intigriti's triage standards based on the proof of concept.

How do I get access to test LangSmith?

Sign-up is open, so you can create an account at https://smith.langchain.com/ with your @intigriti.me email and use it to test. Please only test with accounts and data that belong to you.

Where can I find more information about how to set up and use LangChain's products?

Our documentation at https://docs.langchain.com/ covers how to set up and use LangChain's products.

Do you provide paid plans or licenses for researchers?

Unfortunately, we do not provide paid plans or licenses for testing.