Sanitized live incident
Suspicious activity
Native source identity and targetable endpoints are private.
- Confidence
- 99%
- First seen
- Aug 18, 10:46:54 PM PDT
- Evidence through
- Aug 18, 11:09:58 PM PDT
- AI status
- Complete
Likely true positive for suspicious workload execution: event-driven telemetry directly observed multiple root-context dash shells, a root-context bash child shell, and root-context discovery executables (`id` and `hostname`) in one workload (process evidence [redacted], [redacted], [redacted], [redacted], [redacted]). This supports execution and discovery inside the workload, but not exploitation, remote-request causality, persistence, host escape, or data theft. The origin and authorization remain unresolved because no incident-cited HTTP or flow evidence was available.
- Attack stage
- Execution and discovery within workload; initial access unconfirmed
- Model
- gpt-5.6-sol · 6 evidence calls
Observed impact
- Root-context shell processes executed inside the workload [redacted].
- Root-context discovery processes executed inside the workload [redacted].
- At least one observed root dash lifecycle ended with a nonzero exit; this does not negate that execution occurred [redacted].
Deterministic signals
An event-driven shell execution was observed in a protected workload without correlated HTTP evidence
409 observations · 12 processAn event-driven discovery command was observed in a protected workload without correlated HTTP evidence
3 observations · 3 processA previously correlated process lifecycle exited
411 observations · 12 processExplicit uncertainty
- No incident-cited HTTP event IDs were available. An HTTP lookup using a cited process ID was rejected because that ID is not cited on the HTTP plane, so the initiating request, if any, cannot be determined.
- No incident-cited flow event IDs were available. A flow lookup using a cited process ID was rejected because that ID is not cited on the flow plane, so network consequences and request-to-socket causality cannot be assessed.
- The process summaries expose executable names and lineage but not command arguments or operator intent; authorization and purpose are therefore unknown.
- The recurring parent PID is visible, but its executable, service role, and authorization context are not established by the inspected evidence.
- The source key denotes a workload cluster, not a specific person or remote actor.
- There is no cited evidence proving exploitation, persistence, host escape, lateral movement, command-and-control, or data theft.
Recommended actions
- Immediately identify the owner and executable associated with recurring parent PID 3076121 and compare the activity with deployment, CI/CD, health-check, and administrative change records.
- If the shell and discovery activity is not explicitly authorized, contain the affected workload according to policy, preserve telemetry, and redeploy it from a known-good immutable image.
- Review workload, application, orchestration, and identity-provider logs around 05:46–06:10 UTC to determine the initiating action and principal; preserve relevant records before retention expires.
- Audit the workload image and runtime configuration for unexpected scripts, cron or scheduled tasks, writable executable paths, injected environment configuration, and unauthorized package or file changes.
- Reduce exposure by running the workload as a non-root user where feasible and applying least-privilege capabilities, filesystem, exec, and egress controls.
- Investigate only evidence-supported scope first; do not infer host escape or lateral movement without additional telemetry.