openshell-cli
Warn
Audited by Snyk on Nov 20, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (low risk: 0.10). The required runtime workflows are CLI commands run against the user's own gateway and sandboxes; the only path where outsider-influenced text plausibly reaches the agent is
openshell logsduring Workflow 4's denial monitoring, since denied-connection records include attacker-influenced destination hosts, but an outsider cannot post into that log stream directly and the agent must run the command.
MEDIUM W013: Attempt to modify system services in skill instructions.
- Attempt to modify system services in skill instructions detected (medium risk: 0.30). The skill exposes operational mutations such as gateway registration, global policy and settings changes, provider and profile management, and sandbox deletion. Explicit irreversible loss is stated for
sandbox delete --all, which removes retained state including storage ("Delete remains the operation that removes retained state", "Explicit delete can reclaim a sandbox ... including its storage"), and for suspended-sandbox memory state being discarded. Sandboxes are workspace-scoped resources on a remotely managed gateway, so scope is presumed shared, but the destruction scope is per sandbox. This makes it medium rather than high: no single operation joins irreversible loss with clearly shared infrastructure.
Issues (2)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W013
MEDIUMAttempt to modify system services in skill instructions.
Audit Metadata