Sentrail documentation
Sentrail documentation
Updated August 17, 2026
What Sentrail does
Sentrail is an evidence-gated security layer for AI-built applications. It reads connected systems, runs defined checks, records when each check completed, and produces one of four launch-readiness verdicts.
The basic workflow
- 01
Create a project
Choose a project name and enter the authenticated project workspace. - 02
Connect evidence
Link GitHub and Supabase. Add Vercel when deployment posture is part of your stack. - 03
Run the checks
Collect RLS, advisor, repository, dependency, and deployment evidence from the connected systems. - 04
Review the verdict
Resolve blockers, review medium and low findings, and identify any missing or stale evidence. - 05
Approve fixes
Review the exact proposal and risk before approving a draft remediation pull request.
Browse the documentation
Getting started
Create your first project, connect its stack, and run the initial audit.
Read docsGitHub connection
Install the GitHub App, select repositories, and understand repository checks.
Read docsSupabase connection
Connect through OAuth and audit RLS, policies, advisors, and credentials.
Read docsVercel connection
Link a deployment and inspect environment and protection posture.
Read docsLaunch readiness
Understand the seven checks, evidence freshness, score, and verdict states.
Read docsRemediation
Move from finding to approval, draft PR, and verification.
Read docsMCP and agents
Give Claude Code or Cursor project-scoped, audited Sentrail access.
Read docsSecurity guide
Learn the broader pre-launch security workflow for AI-built apps.
Read docsWhen you need help
Start with the connection page for the provider that is missing evidence. If the connection is healthy but a check fails, preserve the error text and check the latest scan history before reconnecting.
For account or product help, visit Support. For a possible vulnerability in Sentrail itself, follow the responsible disclosure policy.