copaw-ai-assistant
Fail
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: CRITICALREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill recommends several high-risk installation methods that execute unverified code from an external domain. These include piping a script to bash ('curl -fsSL https://copaw.agentscope.io/install.sh | bash') and to PowerShell's Invoke-Expression ('irm https://copaw.agentscope.io/install.ps1 | iex'). These patterns allow for arbitrary command execution on the user's machine.
- [EXTERNAL_DOWNLOADS]: The installation process involves downloading multiple external shell scripts and Python packages from sources that are not part of the trusted vendor infrastructure.
- [COMMAND_EXECUTION]: The skill provides tools for local system interaction, including 'list_files' (listing directory contents) and 'summarize_file' (reading file contents), and mentions an optional 'tool_guard' to monitor risky functions like 'os.system'.
- [INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by ingesting untrusted external data via the 'summarize_url' and 'summarize_file' tools. * Ingestion points: URL content fetching and local file reading. * Boundary markers: None present to distinguish data from instructions. * Capability inventory: Network access via 'httpx'/'requests' and file access via 'open'. * Sanitization: The skill only performs simple truncation and does not filter or sanitize the content for potential injection attacks.
Recommendations
- HIGH: Downloads and executes remote code from: https://copaw.agentscope.io/install.sh - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata