@twing/cli@0.2.29
AI Security Review
scanned 11d ago · by lpm-firewall-aiThe source supports a warning for guarded first-party agent integration setup: a deliberate machine install writes machine-level Claude and OpenCode configuration. The install gate and repository checks constrain that behavior, and no concrete attack was established.
Decision evidence
public snapshot- The package runs a postinstall script, gated to global installation or an explicit machine-setup flag.
- Machine setup writes Claude user-level resolver wiring and an OpenCode plugin, then marks the machine managed.
- The resolver acts only when a session is in a repository with a Twing coordinator configuration.
- The CLI also has an admin-gated, attended setup path that writes repository-local install-enforcement hooks.
- The package-level install hook exits for ordinary project dependency installs and self-installs.
- Test-support uses child processes to create temporary Git fixtures; the inspected excerpt shows no production behavior.
- The design commands are described as on-request operations.
Source & flagged code
7 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgPackage source references child process execution.
dist/opencode-adapter.jsView on unpkg · L12Findings
Affected versions and remediation
This report applies to @twing/cli@0.2.29.
See all versions of @twing/cli for other recorded verdicts.
- Review the evidence and your use of @twing/cli@0.2.29 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.