Back to evidence

Sanitized live incident

Suspicious activity

Native source identity and targetable endpoints are private.

criticalopen
Confidence
99%
First seen
Aug 19, 9:22:21 AM PDT
Evidence through
Aug 19, 12:27:45 PM PDT
AI status
Complete
Likely true positive90% confidence

Verified event-driven process telemetry shows repeated root-run dash shells in the protected workload, including child discovery utilities and a root-run cat process classified as targeting a sensitive file. This strongly supports unauthorized or attack-like command execution with discovery and sensitive-file access activity. However, no HTTP or flow evidence references are available to establish the initiating action, actor, exploit vector, request-to-process causality, network consequence, or whether the activity was authorized administration or lab automation.

Attack stage
Command execution with discovery and sensitive-file access
Model
gpt-5.6-sol · 9 evidence calls

Observed impact

  • Root-level shell and discovery processes executed inside the workload.
  • A root-level cat process classified as targeting a sensitive file executed; actual file contents obtained or disclosed are not established.
  • Observed command processes exited, including verified zero-outcome lifecycles for sampled shell/child chains.

Deterministic signals

Process.observed shell spawn99%

An event-driven shell execution was observed in a protected workload without correlated HTTP evidence

68 observations · 12 process
Process.observed discovery command88%

An event-driven discovery command was observed in a protected workload without correlated HTTP evidence

10 observations · 10 process
Process.correlated exit99%

A previously correlated process lifecycle exited

72 observations · 12 process
Process.observed sensitive file command99%

An event-driven process targeted a sensitive file in a protected workload without correlated HTTP evidence

3 observations · 3 process

Explicit uncertainty

  • No HTTP evidence event is cited by this incident, so the originating request or action, exploit vector, actor, and request-to-process causality cannot be determined.
  • No flow evidence event is cited by this incident, so outbound communication, destination novelty, command-and-control, or exfiltration cannot be assessed.
  • The source key is a workload cluster rather than a guaranteed human or agent identity.
  • The bounded summaries omit command arguments and the sensitive file path; successful content acquisition, modification, or disclosure is therefore not established.
  • The telemetry does not establish whether the commands were unauthorized, legitimate administration, health checks, or intentional attack-lab automation.
  • No evidence establishes host escape, persistence, lateral movement, command-and-control, or data theft.

Recommended actions

  1. Immediately validate whether the root shell, id, env, and sensitive-file cat executions were expected for this workload and timeframe using deployment, job, operator, and change records.
  2. If the activity is not authorized, isolate or replace the workload using established response procedures and preserve relevant process, container, orchestrator, and application logs before teardown.
  3. Review the workload entrypoint and the long-lived parent process associated with the observed shell children to determine the initiating component and whether it can execute untrusted input.
  4. Retrieve surrounding application, ingress, audit, and orchestration telemetry to identify the initial access path; do not infer HTTP causality from process timing alone.
  5. Review least-privilege controls: avoid running the workload as root where feasible, restrict sensitive-file access, apply read-only filesystems, and constrain unnecessary shell/tool availability.
  6. Hunt for additional executions sharing the same workload lineage and for independently evidenced persistence, outbound connections, credential access, or lateral movement.