.png)
.png)
Vulnerability Disclosure Program
Kindo welcomes responsible security research and appreciates reports that help keep our systems and customers safe. If you believe you’ve found a security issue, please report it privately and we’ll work with you to investigate and remediate.
Scope
This VDP applies to all internet-facing systems and applications owned or operated by Kindo.ai, including:
Kindo web properties
Deep Hat web properties
API interfaces
Acquired companies and related companies
Out of Scope
The following are explicitly excluded from this VDP:
Internal systems not accessible from the internet.
Systems of third-party vendors or partners.
Physical security vulnerabilities (e.g., building access).
Denial of Service (DoS) vulnerabilities. While we appreciate reports of potential DoS vulnerabilities, we ask that researchers refrain from testing them against our systems.
Social engineering attacks (e.g., phishing).
Vulnerabilities in third-party libraries or frameworks unless they are uniquely exploitable in our implementation.
The LLM (Deep Hat) product and associated AI/ML model endpoints. Prompt injection, jailbreak attempts, and other LLM-specific attack techniques against Deep Hat are not in scope for this program.
Responsible Disclosure Guidelines
We request that security researchers follow these guidelines when reporting vulnerabilities:
Submit Comprehensive Findings
Limit Testing to Proof Only
Practice Responsible Disclosure
Report Through Official Channels
Maintain Strict Confidentiality
Safe Harbor
Kindo.ai commits to working with security researchers in good faith. If you follow the guidelines outlined in this VDP, we will not initiate legal action against you for accidentally accessing systems or data while conducting security research. This safe harbor applies even if your actions would otherwise violate our terms of service or other agreements.
Response process
Upon receiving a vulnerability report, we will:
1
Acknowledge receipt of the report within 15 business days.
2
Investigate the reported vulnerability and assess its impact. Investigations will take anywhere from 60 to 90 days.
3
If investigations confirm a reported issue, we aim to remediate confirmed vulnerabilities within 60–90 days of validation, depending on complexity and risk.
Rewards and Severity
We evaluate submissions using an internal severity model that aligns with common industry practice:
While estimated CVSS scores from reporters are welcome, Kindo will make the final determination for severity scoring based on our own research and analysis.
Critical (CVSS 9.0–10.0)
Typically $500–$1,000, and not to exceed $1,000 per issue
Unauthenticated RCE on production infrastructure
Full authentication bypass — accessing any account without credentials (e.g., key exposure allowing arbitrary x-user-id/x-org-id spoofing)
Mass cross-tenant data exfiltration (e.g., SQL injection on the multi-tenant PostgreSQL database)
Compromise of encryption keys (decrypts all user secrets) or OAuth tokens at rest
Sandbox escape leading to host-level code execution
Supply chain compromise affecting customer environments
Bulk access to encrypted secrets, API key hashes, or integration credentials across orgs
High (CVSS 7.0–8.9)
Typically $150–$500
Cross-tenant data access — reading another org's conversations, workflows, or dashboard data
Privilege escalation across tenant boundaries
IDOR accessing sensitive data (message content, secrets, integration credentials) of another user, even within the same org
SSRF from MCP tool proxying reaching internal services or cloud metadata endpoints
Authentication bypass with preconditions (e.g., requires valid session + race condition)
DLP bypass exposing high-sensitivity PII (SSNs, credit cards, medical data)
Webhook signature bypass (e.g. HMAC) or direct webhook token enumeration
Backend policy bypass granting unauthorized access to a protected resource
Prompt injection via MCP tool outputs that exfiltrates user data or credentials
Medium (CVSS 4.0–6.9)
Typically $50–$150
Stored/reflected XSS in an authenticated context with limited scope
CSRF on non-critical state-changing actions (e.g., renaming a dashboard)
IDOR accessing non-sensitive data within the same org (e.g., viewing another member's conversation title, not content)
Privilege escalation within the same org (e.g., member → admin of own org)
DLP bypass that leaks low-sensitivity PII (e.g., names, locations) in a single conversation
Denial of service requiring authentication with limited blast radius
MCP tool output leaking non-sensitive internal metadata
Low (CVSS 0.1–3.9)
No monetary reward (may be eligible for swag or public recognition at our discretion).
Information disclosure of non-sensitive metadata (e.g., software versions, internal path names, stack traces in error responses)
Missing security headers with no demonstrated exploit
Self-XSS or issues requiring implausible user interaction chains
Rate limiting gaps on non-sensitive endpoints
Vulnerabilities only exploitable in non-default configurations
