@tangle-network/sandbox-cli@0.21.31
CLI for Tangle Sandbox operations
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe scanner’s credential and command-execution signals are explained by user-invoked CLI features: native Claude authorization followed by Sandbox secret storage, and Solve submission through a deployment-pinned sponsor endpoint. The manifest has no lifecycle hook, and the inspected code does not show credential exfiltration or unauthorized actions.
Decision evidence
public snapshot- The manifest has no install or lifecycle hook; its scripts are for build, development, type checking, verification and tests.
- The Claude credential flow tells the user to authorize in Claude’s native browser and stores the resulting token as a Sandbox secret.
- The CLI parses a token from Claude’s native consent output, and Solve actions are exposed as explicit `run` and `submit` commands.
- The Solve flow pins its sponsor URL and checks that competition metadata matches the trusted deployment.
- The bundle contains a hard-coded sponsor endpoint on a Tailscale hostname; the reviewed code ties it to the Solve flow and validates it against the pinned deployment.
Source & flagged code
8 flaggedThis package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/index.mjsView on unpkgFindings
Affected versions and remediation
This report applies to @tangle-network/sandbox-cli@0.21.31.
See all versions of @tangle-network/sandbox-cli for other recorded verdicts.
- Review the evidence and your use of @tangle-network/sandbox-cli@0.21.31 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.