1panel-websites
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill defines operational boundaries for managing a 1Panel server using documentation and configuration files. No executable scripts, persistence mechanisms, or privilege escalation attempts were found.
- [INDIRECT_PROMPT_INJECTION]: The skill acknowledges an attack surface where data retrieved from 1Panel tools could contain malicious payloads. This is mitigated by explicit operational instructions.
- Ingestion points: Tool results from website listings, application descriptions, and imported documents.
- Boundary markers: Present; the operation contract (references/operation-contract.md) explicitly requires treating results as data, not instructions.
- Capability inventory: The skill uses a scoped set of MCP tools for listing and creating websites, SSLs, and databases.
- Sanitization: The instructions mandate that the agent must not print API keys, database passwords, or raw credential-bearing payloads.
Audit Metadata