How SheerSight works¶
SheerSight has two parts: a dedicated instance that runs for your organization alone, and the console you sign in to. Your security data stays on the instance. The console only receives short summaries so it can show you charts and counts.
The two parts¶
Your tools (Google Workspace, Microsoft 365, AWS, GitHub, ...)
| read-only API access, started by your instance (outbound)
v
+---------------------------------------------+
| Your dedicated instance (single-tenant) |
| - collects audit and security events |
| - runs detections and correlates incidents |
| - stores your full event data |
| - holds your credentials, decrypted |
+---------------------------------------------+
| outbound only: summaries up,
| signed instructions and encrypted credentials down
v
+---------------------------------------------+
| SheerSight console (app.sheersight.io) |
| - charts, counts, source health, audit log |
| - relays encrypted credentials |
| - sends signed, allow-listed instructions |
+---------------------------------------------+
^
| Zero-Trust sign-in
You
Your dedicated instance¶
A hardened server that SheerSight sets up and maintains for your organization only. It:
- collects from your tools using read-only credentials,
- detects threats with rules for each tool, and groups related alerts into incidents,
- triages: alerts go to Slack, where analysts judge them, and known noise is suppressed under strict safety checks,
- keeps your full event data and alert details.
It has no open inbound ports and no public login. It only makes outbound connections: to your tools' APIs, and to the console.
The console¶
The web app at app.sheersight.io, hosted on Cloudflare. It shows what your
instance reports, lets you connect tools, and keeps a tamper-evident audit log.
It never holds your credentials in readable form and never receives your raw
event data.
What travels between them¶
| Direction | What | Protection |
|---|---|---|
| Instance to console | Alert and incident summaries (IDs, rule name, entity, time, disposition), daily triage counts, source health | Allow-listed fields only; anything that looks like a secret is removed |
| Console to instance | Credentials you connect | Encrypted in your browser to your instance's key; the console only ever has ciphertext |
| Console to instance | Instructions (start a source, stop a source, upgrade) | Cryptographically signed; the instance verifies the signature and runs only a fixed list of actions, never a shell |
Raw events, full alert bodies and log contents are not sent to the console.
Connecting a tool, step by step¶
- You create a read-only credential in the tool.
- You paste it into the console. Your browser encrypts it with your instance's public key before sending it.
- The console stores the encrypted blob and tells your instance it's waiting.
- Your instance fetches it, decrypts it locally, and starts collecting.
- Source health flows back to the console, and the card turns live.
Where things are stored¶
| Data | Where | How long |
|---|---|---|
| Raw events from your tools | Your dedicated instance | See Data handling & retention |
| Alerts and incidents, full detail | Your dedicated instance | See Data handling & retention |
| Alert and incident summaries, daily counts | Console | 90 days |
| Encrypted credentials | Console (ciphertext only) and your instance (decrypted) | Until you disconnect or we offboard you |
| Audit log | Console | Kept for the life of your account |