arkcli-helper

Pass

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill manages agent configurations by executing the arkcli binary and modifying local configuration files such as ~/.claude.json, ~/.trae/mcp.json, and ~/.codex/config.toml. These actions are local to the user's environment and are the primary intended purpose of the skill.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of vendor-owned tools, including the @byted-supabase/cli Node.js package and an installation script for the 'Agent Evolve' utility. These resources are hosted on the vendor's official domains and represent standard setup procedures for the Volcengine ecosystem.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user intents to route queries to specific data providers. It implements a security gate (routing.v1) that uses unique turn identifiers and strict provider mapping to prevent unauthorized tool access or data leakage from malicious search results.
  • [SAFE]: The skill handles sensitive API keys by writing them to .env files with restricted filesystem permissions (0600). The instructions explicitly forbid the agent from printing keys or passing them manually in conversation, adhering to security best practices for secret management.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 17, 2026, 12:11 PM