arkcli-profile

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill logic involves reading identifiers such as profile names and resource IDs from tool outputs and using them as arguments in subsequent CLI commands. This pattern represents an ingestion surface for potentially untrusted data originating from local configuration files.\n
  • Ingestion points: Profile names and resource IDs are retrieved via arkcli profile list and arkcli resources list in SKILL.md.\n
  • Boundary markers: The instructions lack explicit directives for the agent to treat these identifiers as potentially untrusted strings or to apply escaping.\n
  • Capability inventory: The skill uses arkcli for subprocess operations including creation, deletion, and modification of local configuration files.\n
  • Sanitization: No explicit sanitization steps are provided in the instructions for the agent when processing these identifiers.\n- [COMMAND_EXECUTION]: The skill's primary function is the execution of CLI commands via the arkcli binary. While these commands are directed at a vendor-specific tool, they involve the dynamic assembly of shell commands based on user and system-provided inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 04:04 PM
Security Audit — agent-trust-hub — arkcli-profile