skills/unicorn-plugins/abra/setup/Gen Agent Trust Hub

setup

Fail

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple high-impact system commands including git clone, docker compose, pip install, and executing local Python scripts. These operations significantly alter the host environment.
  • [COMMAND_EXECUTION]: In Phase 12, the skill modifies the agent's internal configuration file (.claude/settings.local.json) to inject a broad set of permissions (Read, Write, Edit, Bash) for the {ABRA_PLUGIN_DIR}. This effectively automates permission expansion, bypassing the usual manual review process for these specific capabilities.
  • [EXTERNAL_DOWNLOADS]: The skill fetches external content from github.com/langgenius/dify.git and interacts with marketplace.dify.ai to download and install plugins. While these are well-known services associated with the Dify project, they involve running unverified third-party code.
  • [CREDENTIALS_UNSAFE]: The skill explicitly collects and handles sensitive secrets, including DIFY_PASSWORD and GROQ_API_KEY, storing them in .env files. While this follows common developer practices, the automated handling and storage of plaintext credentials in accessible files increases the risk of exposure.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 23, 2026, 02:09 PM
Security Audit — agent-trust-hub — setup