tailscale
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and follow instructions from an external local file located at
~/docs/devices/tailscale.mdfor connection steps, supported commands, and capability definitions. - Ingestion points:
~/docs/devices/tailscale.mdreferenced inSKILL.md. - Boundary markers: None present to distinguish instructions in the documentation from system-level constraints.
- Capability inventory: The skill uses the
/opt/hatch/bin/tailscalebinary for network egress, VPN configuration, and TCP tunnel proxying. - Sanitization: No evidence of sanitization or validation for the content read from the documentation file.
- [COMMAND_EXECUTION]: The instructions explicitly direct the agent to use a bundled command-line utility at
/opt/hatch/bin/tailscaleto manage networks, join tailnets, and reach private machines.
Audit Metadata