terminal-smart-cli@0.97.68
Terminal Smart no seu terminal — pergunte, analise logs por pipe e orquestre agentes de IA. Comando: ts
AI Security Review
scanned 4d ago · by lpm-firewall-aiThe manifest has no install hook, and the CLI's network and credential flows are tied to user-invoked features and the package service or a user-configured AI provider. The scanner's browser-credential and unofficial-gateway claims were not substantiated in the reviewed source.
Decision evidence
public snapshot- The manifest exposes a user-invoked `ts` CLI and declares no install lifecycle scripts.
- The CLI dispatches explicit commands such as login, chat, agent, and video; an empty invocation shows help without a saved token.
- The keyring uses a configured provider key and base URL when BYOK is active; otherwise it retrieves the package service key through the API.
- The vision helper sends prompts and images to the resolved AI endpoint as part of its visual-check feature.
- The default backend URL is the package's documented service, `https://terminalsmart.com.br`.
- No source evidence reviewed showed browser credential collection or a fixed unofficial credential redirect.
- The CLI has process execution, filesystem, and network capabilities, including project analysis and browser-based visual checks.
- A configured BYOK key is sent to the corresponding configured provider endpoint when the user invokes AI features.
Source & flagged code
9 flaggedThis package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
bin/ts.jsView on unpkgFindings
Affected versions and remediation
This report applies to terminal-smart-cli@0.97.68.
See all versions of terminal-smart-cli for other recorded verdicts.
- Review the evidence and your use of terminal-smart-cli@0.97.68 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.