Skills
skills/auto-blog-writer-app/autoblogwriter-skill/autoblogwriter-cli/Gen Agent Trust Hub

autoblogwriter-cli

Fail

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The README.md file provides installation instructions that involve piping a script from 'https://skills.sh/install' directly to bash. This practice enables arbitrary remote code execution from an unverified source.
  • [EXTERNAL_DOWNLOADS]: The skill directs the download of installation scripts and skill configuration from the untrusted domain 'skills.sh'.
  • [COMMAND_EXECUTION]: The skill is built around executing multiple autoblogwriter CLI commands, which allows the agent to perform actions such as generating ideas, creating posts, and running workflows on the host system.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection: 1. Ingestion points: The skill processes external content from payload.json, workflow.json, and user-supplied --inline JSON strings. 2. Boundary markers: There are no explicit markers or delimiters to isolate untrusted data from instructions. 3. Capability inventory: The agent can execute mutating CLI commands, including publishing posts and starting run workflows. 4. Sanitization: No sanitization or input validation is specified for data used within CLI command arguments.
Recommendations
  • HIGH: Downloads and executes remote code from: https://skills.sh/install - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Mar 9, 2026, 07:32 AM