In short: if you find a security vulnerability in Merget, email security@merget.ai. We acknowledge every report within 3 business days, we fix what you found, and we will not take legal action against research that follows this policy. We do not pay bounties. We ask you to keep the details private until a fix ships or 90 days have passed. A machine-readable copy of this policy's contact lives at https://merget.ai/.well-known/security.txt.
1. Scope
This policy covers the Merget merge queue and the sites around it:
- merget.ai, the website;
- sema.merget.ai, the dashboard, the HTTP API under
/v1and the MCP server; - auth.merget.ai, sign-in, two-factor authentication and agent consent;
- demo.merget.ai, the demo;
- the Merget GitHub App: what it reads, what it writes and how it is authorized;
- the Merget Claude Code plugin, its skills and the MCP server they connect to;
- the Merget CLI;
- the build and validation environment in which Merget runs a repository's own setup, build and test commands, and the isolation between repositories there;
- the emails and text messages Merget sends to sign you in.
We are most interested in: access to another organization's or repository's data; escaping the build environment, reaching private network addresses from it, or obtaining Merget's own credentials from it; an agent doing something its permissions or its organization's settings refuse; bypassing two-factor sign-in, token revocation or agent revocation; writes to a repository that the automation mode does not allow, including anything under .github/; validation secrets reaching a place the documentation says they never reach; and session recordings carrying text the dashboard or the website says it masks.
2. Out of scope
Please do not report, and do not test for:
- denial of service, rate-limit exhaustion, or anything that degrades the Service for other customers;
- social engineering, phishing or physical attacks on Merget's people, customers or providers;
- vulnerabilities in third-party services Merget uses (GitHub, Google Cloud, Stripe, PostHog, Cal.com, Cloudflare, Slack): report those to the vendor under its own policy;
- findings in a customer's own repositories, pull requests or CI, including findings Merget's analysis reports about their code;
- missing best-practice headers, DNS records or TLS settings with no demonstrated impact, version disclosure, clickjacking on pages with no state-changing action, and reports from automated scanners with no proof of exploitability;
- the demo's simulated data, which is not real;
- a vulnerability that requires a compromised device, a malicious browser extension or physical access to a signed-in session.
3. How to report
Email security@merget.ai. Write in English. Include:
- what you found and where (the host, the route, the tool, the page, or the file in the plugin), and the account or organization you used, if any;
- the steps to reproduce it, with requests and responses where they help;
- the impact as you understand it: what an attacker could read, change or do;
- whether anyone else knows, and whether the details are published anywhere;
- how we can reach you for questions, and the name to credit if you want credit.
Do not send another customer's data. If you reached it by accident, stop, tell us what you saw, and delete your copy. If the report concerns the plugin repository, you may also open a private security advisory there; the SECURITY.md file in that repository points back to this policy.
We do not sign reports today, so the policy is published without a PGP key. If you need to send something that must not travel in plain email, say so in a first message and we will arrange a channel.
4. What happens next
- Acknowledgment within 3 business days. You get a reply that we received the report, with a reference to quote.
- Triage. We confirm the issue, rate its severity, and tell you what we found and what we plan to do. We may ask you for more detail.
- Fix. We fix confirmed issues in priority order and tell you when the fix ships. Where a fix affects customers' data or requires them to act, we tell the affected organizations' Owners without undue delay, as the Privacy Statement says.
- Credit. With your agreement, we credit you by name when we announce the fix.
5. Rules for research
Research is covered by this policy when you:
- test only accounts, organizations and repositories you own or have permission to use, and never another customer's;
- stop as soon as you have enough to show the issue, and do not read, change or keep more data than that;
- do not degrade the Service, run denial-of-service tests, or spam the queue with pull requests;
- do not use social engineering, phishing or physical attacks;
- do not exploit an issue beyond what the report needs, and do not pivot from one finding into other systems;
- keep what you found private until the coordinated disclosure in section 8 allows publication;
- comply with the law, and with the Terms of Service and the Acceptable Use Policy except where this policy says otherwise.
6. Safe harbor
Research that follows this policy is authorized. For such research Merget will not pursue civil action, will not refer you for prosecution, and will not bring a claim under the Computer Fraud and Abuse Act, the Digital Millennium Copyright Act or similar laws for circumventing the technical measures that research required. Where your research is conducted in good faith but accidentally breaks a rule above, tell us at once; we will work it out with you rather than against you. If a third party brings an action against you for research that followed this policy, we will say that your research was authorized.
This safe harbor covers Merget's own systems. It cannot authorize testing against third parties' systems, and it does not cover anyone who was not acting in good faith.
7. No bounty
Merget does not run a bug bounty and does not pay for reports. We thank researchers and credit them when they want credit.
8. Coordinated disclosure
We ask you not to publish a vulnerability until a fix has shipped, or 90 days have passed since our acknowledgment, whichever comes first. If a fix needs longer, we will tell you why and ask for more time; if customers must act to be safe, we may publish sooner, with you. Before publishing, let us read the draft, so that we can check it for other customers' data and agree on the date.
9. Changes
We may change this policy. A change takes effect when published; a report sent under the policy in force at the time is handled under it. The Expires field in https://merget.ai/.well-known/security.txt says when the published contact information must be checked again.
Contact
Merget, Inc.
418 Broadway, Unit 7259, Albany, NY 12207
security@merget.ai for vulnerability reports; support@merget.ai for everything else