uv-package-manager
Fail
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to download the uv installer and components from the official developer domain at astral.sh and the project's GitHub repository.- [REMOTE_CODE_EXECUTION]: Installation steps include piping remote scripts directly into the shell (curl ... | sh and irm ... | iex), which is a common but sensitive execution pattern for remote code.- [COMMAND_EXECUTION]: The skill demonstrates multiple shell commands for managing projects, environments, and executing Python code via the uv run command.- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection (Category 8). It ingests untrusted data from project files such as pyproject.toml, uv.lock, and requirements.txt (Ingestion points in SKILL.md). No explicit boundary markers or sanitization logic are defined to mitigate malicious instructions within these files. The agent has the capability to execute commands and modify the file system based on these inputs (Capability inventory: uv run, uv sync, uv add).
Recommendations
- HIGH: Downloads and executes remote code from: https://astral.sh/uv/install.sh - DO NOT USE without thorough review
Audit Metadata