repo-scan

Warn

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to download its functional code from a community GitHub repository (https://github.com/haibindev/repo-scan.git) using a specific commit hash during installation.
  • [COMMAND_EXECUTION]: The installation section requires the execution of multiple shell commands (mkdir, git init, git remote, git fetch, git checkout, cp) to configure the local environment and set up the skill.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core function involves scanning and auditing untrusted third-party source code repositories.
  • Ingestion points: The skill reads files across diverse stacks (C++, Android, iOS, Web) within the user's repository during the scanning process.
  • Boundary markers: The documentation does not mention the use of delimiters or specific instructions to the agent to disregard natural language commands found within the scanned source code.
  • Capability inventory: The skill reads local files and generates interactive HTML reports based on the analysis of those files.
  • Sanitization: There is no mention of sanitizing or validating the content of the scanned files to prevent embedded instructions from influencing the agent's output.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 6, 2026, 04:01 AM
Security Audit — agent-trust-hub — repo-scan