ru
Fail
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The installation section of the documentation provides a command to download and execute a Bash script directly from a remote GitHub repository (
https://raw.githubusercontent.com/Dicklesworthstone/repo_updater/main/install.sh | bash). This practice is a security risk as it executes remote content in the local shell environment. Since the source is the author's own repository, it is documented as the standard installation path for this tool. - [INDIRECT_PROMPT_INJECTION]: The skill's
agent-sweepand review orchestration features ingest untrusted data from GitHub (issues, pull requests) and local repository files to be processed by AI agents. This creates a surface where malicious instructions hidden in these sources could potentially influence the agent's actions during the automated review or commit process. - Ingestion points: GitHub Issues, Pull Requests, and file contents in managed repositories (as seen in
SKILL.md). - Boundary markers: The skill does not explicitly detail the use of delimiters to isolate untrusted data in its prompt instructions, though it mentions denylists.
- Capability inventory: The tool performs Git operations, network calls to GitHub, and subprocess execution via integrated AI sessions.
- Sanitization: The tool implements file denylists and secret scanning (
--secret-scan) to prevent sensitive data from being staged or committed. - [COMMAND_EXECUTION]: The tool is a Bash CLI that executes a wide range of Git plumbing commands and orchestrates terminal multiplexer (tmux) sessions for interactive AI agent workflows.
Recommendations
- HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/Dicklesworthstone/repo_updater/main/install.sh - DO NOT USE without thorough review
Audit Metadata