Skills
Navigate
skills/alphaxiv/openresearch/orx-compute/Gen Agent Trust Hub

orx-compute

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTIONPRIVILEGE_ESCALATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill reads and follows instructions from an external file (CUSTOM.md) via the orx compute instructions show command.\n
  • Ingestion points: The content of CUSTOM.md is loaded into the agent's context as shown in SKILL.md and references/configuration.md.\n
  • Boundary markers: No specific delimiters are used to isolate these instructions from the core skill instructions.\n
  • Capability inventory: The skill can execute shell commands, modify SSH configurations, and interact with cloud provider APIs.\n
  • Sanitization: Instructions from CUSTOM.md are not sanitized or validated before being presented to the agent.\n- [DYNAMIC_EXECUTION]: The Kubernetes backend reference in references/k8s.md utilizes eval "$ORX_SCRIPT" to run workloads inside pods, representing dynamic shell execution of environment-variable-stored scripts.\n- [COMMAND_EXECUTION]: The skill relies on executing multiple subcommands of the orx CLI tool to orchestrate compute resources.\n- [PRIVILEGE_ESCALATION]: The SSH backend reference (references/ssh.md) describes the automated modification of the user's ~/.ssh/config file to manage connection multiplexing, which is a persistent modification of a sensitive system configuration file.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 02:23 PM