Skip to content

Investigations

This page covers Investigations in the hosted dashboard, part of Omnodex Pro. One finding tells you one call looked risky. An investigation tells you a story: an agent read a credential, packed it into an archive, and sent it to a host this machine had never reached. Each step alone might be routine. Together they’re worth a look.

Advanced rules follow four kinds of sensitive data through a session: credentials and secrets, database files and dumps, mail stores, and source code. They track each one from where it was read, through any archiving or encoding, to where it was sent. Findings about the same chain are grouped as one investigation.

  • Your weekly review. Open the page, sort by severity, and work down the list. Most chains are routine; the page is built so you can tell which in a minute or two.
  • After a notification. A notification about a data movement finding links to the finding; its investigation shows the whole chain around it.
  • When someone asks “did an agent send our code anywhere?” Filter by rule or destination across every stream and machine on the account.
  1. Open Investigations in the sidebar. Every investigation across your sessions is listed, Most severe first, with its severity, the number of steps and findings, the destination and when it was last seen. Switch the sort to Newest first to see what’s recent.

  2. Narrow the list. Use the time range and the filters for severity, stream, machine, session, destination and rule. The time range is shared with the Dashboard and Activity pages.

  3. Open one. Each call in the chain is listed in order with its role: where the data came from, what was done to it, or where it went. Each shows what the call did and the call behind it.

  4. Check the destination. The investigation names it and says whether this machine had reached it before. first time reached is the strongest signal on the page: an agent sending sensitive data somewhere new.

  5. Read why it might be routine. Each investigation lists reasons the chain could be normal work, such as a credential sent as authentication to its own service. Credentials are shown by kind, and whether they were sent to authenticate or inside a request, never by value.

  6. Decide. If it was expected, choose Allow like this and scope it to that host and project, so the same pattern stops raising findings. The entry goes to your Policies. If it wasn’t, Open this session on the dashboard takes you to the session and machine to look into, and the credential kind tells you what to rotate.

The calls and their order are observed: Omnodex recorded each one. That the data read is the data sent is inferred, because Omnodex sees the calls, not the bytes they moved. An investigation is a reason to look, not proof that data left.

For activity through the MCP proxy, as the last call in the chain is recorded. For activity captured by agent hooks, when the session ends or at the next background pass, within about 15 minutes. Either way, the calls have already run. To stop a call before it runs, use guardrails.

Investigations also appear at the top of the Risk Events panel on the Dashboard, for the sessions in view, most severe first.

The findings behind investigations come from advanced rules, which are part of Pro. On other plans they’re held back: the page says investigations come with Pro, the Dashboard shows a line below the connection graph saying how many were found, and sessions are scored only from the findings shown. Each rule is explained in the Risk Library.