unifi-protect-support
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides a link to documentation on the vendor's GitHub repository ('https://github.com/sirkirby/unifi-mcp/blob/main/docs/support-bundles.md'). This serves as an authoritative resource for users seeking canonical guidance on the support process.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data returned from the 'protect_get_support_bundle' tool, which creates a surface for ingesting information from the monitored environment.
- Ingestion points: Data returned by the 'protect_get_support_bundle' tool execution specified in 'SKILL.md'.
- Boundary markers: The instructions explicitly mandate showing the full JSON output to the user and pausing for manual review and confirmation before any external sharing occurs.
- Capability inventory: The skill is focused on data formatting and display; it does not request tools for network exfiltration, arbitrary command execution, or file system modification.
- Sanitization: The documentation states that the tool output is sanitized by stripping credentials, identifiers, and raw configuration files before being provided to the agent.
Audit Metadata