agent-publisher-admin

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides a CLI tool in scripts/agent_publisher_admin.py for administrative tasks. The commands allow users to interact with a backend service to manage agents, styles, and AI tasks via user-initiated shell commands.
  • [CREDENTIALS_UNSAFE]: The script implements token-based authentication. It stores the resulting token in ~/.agent-publisher-token with restricted file permissions (0600) to ensure only the current user can read the file, which is a standard security practice for CLI tools.
  • [EXTERNAL_DOWNLOADS]: The script communicates with a backend API using the httpx library. The backend URL is configurable via the AP_URL environment variable, defaulting to a local address, facilitating legitimate integration with the user's infrastructure.
  • [DATA_EXFILTRATION]: The CLI includes a feature to read content from a local file via the --prompt-file argument for uploading custom prompts to the backend. This behavior is documented and aligns with the administrative purpose of managing style presets.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 08:20 PM
Security Audit — agent-trust-hub — agent-publisher-admin