Skills
skills/agentscope-ai/copaw/file_reader/Gen Agent Trust Hub

file_reader

Pass

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes execute_shell_command to run file and tail utilities. Evidence: Commands found in SKILL.md. Risk: If user-supplied filenames are not correctly escaped, it could lead to command injection.
  • [PROMPT_INJECTION]: The skill reads external content, creating a surface for indirect prompt injection. Ingestion points: read_file and tail operations on files. Boundary markers: No delimiters or warnings are used to sequester file content. Capability inventory: Access to file reading and shell command execution. Sanitization: File content is summarized without prior validation or sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 4, 2026, 04:09 PM