Trust & information

Security

Understand Gitflare's authorization boundaries and report security concerns.

Updated October 18, 2026 · nexarq AI (nexarq Technologies Pvt Ltd)

Access follows your permissions

Workspace roles and resource grants limit access. Connected MCP clients receive scoped permissions that remain limited by the authorizing member's role. Task credentials are separately scoped and cannot authorize personal AI clients. Review active sessions and grants in your account settings and revoke access you no longer need.

Code, credentials and changes

Coding agents work in isolated environments. Canonical repository and deployment credentials are kept outside task coding-agent sandboxes. Cloudflare connection tokens are stored encrypted. Approval is tied to the candidate revision and checked state; review the change and destination before landing or deploying.

These controls reduce risk but do not establish a guarantee that code, dependencies or generated output are secure. Keep secrets out of repository files and prompts, restrict provider permissions and retain appropriate backups.

Report a vulnerability

Email security@gitflare.io with the affected page or component, a clear description, reproduction steps and the impact you observed. Share only the minimum evidence needed and redact credentials and personal information. Use private email for vulnerabilities rather than posting exploitable details in public issues.

Do not access other users' data, disrupt the service or continue testing after you have demonstrated the issue. We have not published a guaranteed response time or a bug bounty reward commitment.