How we secure SentinelPanda
A compliance product has to clear the bar it asks customers to clear. The commitments here apply to every tenant, regardless of plan, and they describe how we protect customer data, how we run our own compliance program, and how to report a vulnerability.
Six commitments
A compliance product has to clear the bar it asks customers to clear. The commitments below are non-negotiable and apply to every tenant, regardless of plan.
- HMAC-signed audit log. Every state change, comment, and approval is written to an append-only log, and each entry carries its own HMAC signature so tampering is detectable.
- AES-256-GCM credentials. Integration secrets are encrypted at rest and decrypted only inside the sync worker, at the moment a sync runs.
- SSO + MFA (org-enforceable). Per-tenant OIDC single sign-on, or email + password — with TOTP multi-factor authentication your organization can require, plus an idle session timeout.
- Per-tenant isolation. Every record is scoped to a tenant and access is enforced server-side on every call — never at render time.
- Three-layer RBAC. Role, tenant, and resource. Permissions are checked on every server call.
- QSA + auditor visibility. External auditors and your QSA join your workspace with auditor-layer seats. Every approval, rejection, and request-for-info is recorded to the HMAC-signed audit log.
How we protect customer data
Customer content is encrypted in transit (TLS 1.2+) and at rest. Logical access is gated by SSO with org-enforceable MFA. When SentinelPanda staff open a customer workspace, the access is recorded in that customer's own audit log, and staff accounts require MFA. SentinelPanda is built and run by one person, so a change is not reviewed by a second engineer. Instead, every change goes through a pull request with a written changelog entry and must pass the automated checks described below before it is merged. There is one production environment; development and automated testing run separately, on synthetic data.
Malware scanning
Every file uploaded as evidence is scanned for malware before it is stored. If the scanner is unavailable, the upload is refused rather than accepted unscanned. Downloads are always served as attachments, never rendered in the browser.
Security testing
Every change to the product runs through automated checks before it is merged: the full test suite, static analysis of the backend code, dependency vulnerability audits, secret scanning, and a scan of dependencies and source files for known vulnerabilities. The container image is scanned before a release. Tenant isolation is tested on every change by a suite that attempts to reach one tenant's data from another across every API route. An independent penetration test has not been carried out yet; when one is, its executive summary will be published on this page.
AI features and your data
AI-assisted features (control-mapping suggestions and questionnaire answer drafts) send only the text needed for the task to Anthropic's API, under commercial terms that do not allow customer inputs or outputs to be used for training models. AI output is always a draft: a person reviews and accepts it before it becomes part of your record. AI-assisted features can be switched off for a workspace on request.
Data processing agreement
Customers can sign a data processing agreement covering GDPR Article 28 and UK GDPR obligations, including our sub-processors, security measures, breach notification and deletion at the end of service. Request it at security@sentinelpanda.com.
How we run our own compliance
We use the product to track our own controls. SentinelPanda does not hold a SOC 2 report or an ISO 27001 certificate today, and no auditor has been engaged yet. When that changes, the report type, the period it covers and the auditor will be published here.
Reporting a vulnerability
If you believe you have found a security vulnerability, please report it through our responsible disclosure programme. We respond to every report we receive.
Contact
Security questions: security@sentinelpanda.com. A signed PGP version of this address is published in /.well-known/security.txt.