Responsible Disclosure

Responsible Disclosure Program

Baponi is a sandboxed code execution platform for AI agents. We welcome security researchers to test our isolation boundaries and report findings through this program.

Baponi's sandbox isolation is the core of our platform. We invest heavily in defense-in-depth: namespace isolation, syscall filtering, capability dropping, network isolation, credential separation, and immutable audit logging. Every layer is enforced at the kernel level. The entire stack is written in Rust.

We welcome security researchers to test these boundaries. If you can demonstrate a structural escape from our isolation architecture, we want to know -and we'll compensate you for the work.

Scope

What qualifies for a reward

Qualifying findings are eligible for a reward of up to $10,000 USD, depending on severity and impact.

Sandbox escape

Arbitrary code execution on the host or outside the sandbox boundary, originating from within a standard Baponi execution environment (Python, Node.js, or Bash).

Cross-tenant data access

Demonstrable ability to read, modify, or exfiltrate data belonging to another tenant's isolated sandbox from within your own sandbox environment.

Out of scope

  • Denial of service (DoS/DDoS), resource exhaustion, or actions that degrade platform availability
  • Attacks against Baponi's web application, API gateway, or admin console (XSS, CSRF, SSRF, SQLi) -these are standard application vulnerabilities, not sandbox isolation issues
  • Social engineering, phishing, or physical attacks against Baponi employees or infrastructure
  • Attacks against third-party services, cloud provider infrastructure, or upstream dependencies
  • Automated scanning or fuzzing that generates excessive load
  • Vulnerabilities in software not developed or maintained by Baponi
Rules

Rules of engagement

Follow these rules to ensure your research qualifies for the bounty and is protected by our safe harbor.

1

Use your own account

All testing must be conducted using your own Baponi account on the free tier. Do not target, access, or attempt to access data belonging to any other user.

2

Register a Research Identifier

Before beginning any testing, contact us to obtain a unique Research Identifier. Include this identifier in all sandbox execution requests via the X-Research-Id HTTP header. This allows us to distinguish authorized security research from malicious activity, ensure safe harbor protections apply to your testing, and correlate your findings with observed activity during triage.

3

Prove it on the live platform

Your proof of concept must be fully reproducible on Baponi's live production environment at the time of submission. Theoretical vulnerabilities, local-only exploits, or attacks that work in other environments but fail on ours do not qualify.

4

Bypass our isolation, not upstream software

Qualifying vulnerabilities must demonstrate a bypass of Baponi's isolation architecture -the security layers we designed and maintain. Exploits that rely solely on unpatched vulnerabilities in third-party software (Linux kernel, container runtime, cloud provider infrastructure) without bypassing any Baponi-specific isolation layer do not qualify independently. If a known vulnerability is used as one step in a chain that also bypasses our isolation layers, the full chain qualifies.

5

One bounty per root cause

Multiple manifestations of the same underlying vulnerability are treated as a single finding. The bounty is awarded to the first researcher to submit a valid, reproducible proof of concept for a given root cause.

6

No disruption

Do not degrade platform performance, destroy data, or interrupt service for other users. If you believe testing a vulnerability may cause disruption, describe the approach in your report and we will work with you to test it safely.

Timeline

What to expect after you submit

Acknowledgment

Within 72 hours

We confirm receipt of your report.

Triage

Within 7 business days

We assess severity, reproduce the finding, and determine eligibility.

Resolution

Varies by severity

We develop and deploy a fix. We keep you informed throughout.

Reward

Within 30 days of validation

Reward paid after we confirm the finding qualifies under program rules.

Disclosure

Coordinated

We agree on a disclosure timeline after the fix is deployed.

Safe Harbor

Our commitment to researchers

Baponi will not pursue legal action against security researchers who conduct testing in compliance with this policy. We consider authorized security research conducted under this program to be:

  • Authorized under the Computer Fraud and Abuse Act (CFAA) and equivalent international laws
  • Exempt from DMCA restrictions on circumventing technological protection measures
  • Conducted in good faith and within the boundaries of this program

If a third party initiates action against you for research conducted in compliance with this policy, we will make it known that your actions were authorized.

Disclosure: All submissions must be reported privately through our contact page. Public disclosure before we have deployed a fix voids the bounty and safe harbor protections. We will coordinate a mutually agreed disclosure timeline after remediation.

Submit a report

Report vulnerabilities through our contact page. Select "Security" as the topic. Include a clear description, step-by-step reproduction instructions, your Research Identifier, and any supporting evidence.