skills/exmergo/dex/transform/Gen Agent Trust Hub

transform

Pass

Audited by Gen Agent Trust Hub on Nov 2, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The wrapper script scripts/run.py uses uv to fetch and install the exmergo-dex-core package from PyPI. The version is pinned to '1.16.6' and the package originates from the skill's author ('exmergo'). This is the intended delivery mechanism for the skill's core logic.
  • [COMMAND_EXECUTION]: The skill requires the agent to execute shell commands using the uv tool to run the provided scripts/run.py wrapper. The script uses os.execvpe and subprocess.run to manage the engine's execution environment. The wrapper includes a security feature that explicitly filters sensitive environment variables (e.g., GITHUB_TOKEN, CLICKHOUSE_CLOUD_API_KEY) to prevent them from being inherited by the child process.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data including warehouse schemas, SQL files, and ticket/issue text. It contains robust instructions for the agent to treat this content as data only. Specifically, it mandates that no text within processed data can authorize spending increases or confirm billed operations, and requires human intervention for PII flags.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 2, 2026, 04:51 AM