highexperimentalLinuxOCIT1552.001

Linux Agent Accessing OCI Security Token Or API Material

Detects AI agent processes reading OCI session tokens, security credentials, or API key files, indicating potential identity theft or privilege escalation driven by an agent operating outside its authorized scope.

Updated Jan 10, 2025 · OCI AI Security Team

agentic-aicredential-accessoci-tokenidentity-abuselinuxociowasp-asi03

Problem Statement

OCI session tokens and API keys grant access to cloud resources and services. An AI agent reading these credentials beyond the scope of its initial authentication represents identity abuse and could be used to perform unauthorized OCI operations under the agent's cloud identity.

Sample Logs

{"timestamp":"2025-01-10T11:02:18Z","computer_name":"oci-worker-10","user":"agent_svc","image":"/usr/bin/python3","target_filename":"/home/agent_svc/.oci/sessions/DEFAULT/security_token","access_type":"read"}

Required Fields

image

target_filename

user

computer_name

False Positives

·OCI SDK initialization code that reads token files at startup for session-based authentication
·Token refresh daemons that periodically update session tokens

Tuning Guidance

Baseline expected token reads at agent startup and suppress repeating patterns. Alert on reads from processes not in the approved agent binary list or reads occurring long after initialization.