No access to your systems
- Vigil never connects to your clusters, cloud accounts, logs or monitoring tools.
- No credentials, agents or API keys from you are needed or accepted.
- Vigil cannot run commands or make changes. Engineers decide what to run.
Security
Built for teams who can’t afford surprises.
Vigil helps you investigate without ever touching production. Here is exactly what happens to your data, and how you can check it yourself.
- No access to your systemsNo agents, credentials or connections
- No stored conversationsUnless you choose to share one
- Secrets redacted twiceIn your browser, then on our server
- Locked-down browserHSTS, strict CSP, no embedding
- Abuse limitsRate limits that fail closed
Data journey
What happens to a message you send.
Four hops, each with its own protection. Select a step to see what it does.
Your browser
On your device
Vigil API
Cloudflare
AI gateway
Private service
AI model service
Private, hosted on AWS
Step 1 · Your browser
- Reads attached logs locally; files are never uploaded
- Redacts keys, tokens and passwords before sending
- Keeps chat history on your device only
Practices
The details.
Your data
- Conversations are not stored on our servers unless you choose to share one.
- Chat history lives only in your browser; rename, delete or clear it anytime.
- Shared chats are opt-in, unguessable, hidden from search, exclude attached files, expire after 30 days and can be revoked.
- Our AI model service does not use prompts or answers to train models.
Safe answers by design
- Read-only checks come first; any step that changes state is clearly flagged.
- Pasted and attached logs are treated as untrusted data, not instructions, to resist prompt injection.
- Messages that look like credentials are rejected; secrets inside attached logs are redacted.
- Hypotheses are labelled as such and tied to the evidence that would confirm them.
Application protections
- HTTPS everywhere with HSTS, a strict Content Security Policy and no embedding in other sites.
- API requests are same-origin only, size-limited and strictly validated.
- Per-visitor and global rate limits; the service fails closed if limits cannot be checked.
- Secrets live in Cloudflare’s encrypted settings and never reach the browser.
Don’t take our word for it
Verify it yourself.
These are the security headers every page is served with. Run the command, or open your browser’s developer tools, to see them on any page.
$ curl -sI https://ai-results-studio-d.pages.dev/
strict-transport-security: max-age=63072000; includeSubDomains
content-security-policy: default-src 'self'; script-src 'self' 'unsafe-inline';
style-src 'self' 'unsafe-inline'; img-src 'self' data:; media-src 'self';
connect-src 'self'; frame-ancestors 'none'; base-uri 'self'; object-src 'none'
x-frame-options: DENY
x-content-type-options: nosniff
referrer-policy: strict-origin-when-cross-origin
permissions-policy: camera=(), microphone=(), geolocation=(), payment=(), usb=()
cross-origin-opener-policy: same-originShared responsibility
Safety is a team effort.
What Vigil does
- Never touches your infrastructure
- Redacts detectable secrets before and after sending
- Puts read-only checks first and flags risky steps
- Stores nothing unless you share a chat
What your team does
- Remove customer and personal data before sharing
- Review every command before running it
- Share links only with the people who need them
- Make the final call on every change
Report a vulnerability
Found a security issue? Tell us privately first. Send a short description through the contact form with “Security” at the start of the note. Please don’t include working exploits or other people’s data.
Early access
Security questions from your team?
Join the waitlist and tell us what your security review needs. Private deployment in your own AWS account is on the Enterprise roadmap.