arkcli-shared

Pass

Audited by Gen Agent Trust Hub on Sep 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill implements a standardized protocol for executing the arkcli binary, requiring specific attribution environment variables (ARKCLI_CALLER_TYPE, ARKCLI_CALLER_NAME, ARKCLI_SKILL_NAME) to be set for every tool invocation to ensure consistent versioning and logging.
  • [EXTERNAL_DOWNLOADS]: The skill documentation includes instructions for installing and updating the @volcengine/ark-cli package via the NPM registry. It also handles the installation of the arkexp experience tool from the vendor's CDN. As these resources belong to the recognized vendor, they are considered standard operational functionality.
  • [INDIRECT_PROMPT_INJECTION]: The skill functions as an entry point and router for instructions that process user input to generate CLI commands.
  • Ingestion points: User intent strings provided in natural language and the structured outputs (JSON/YAML) from arkcli subcommands.
  • Boundary markers: The skill relies on modular sub-skills for specific task execution, and while it mandates structured output parsing, it does not explicitly define prompt delimiters in this shared context.
  • Capability inventory: Access to the full range of arkcli functions, including account authentication, resource deployment, and profile management.
  • Sanitization: The skill explicitly forbids manual text-based parsing (e.g., using sed or head) and requires the use of structured parsers like jq or yq to process tool output, which serves as a mitigation against schema-based injection attacks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 24, 2026, 05:37 AM
Security Audit — agent-trust-hub — arkcli-shared